2 22

etroit,Michigan,USA ct.9-13

IMPORTANT DATES Call for Papers

January 15, 2022
Digest submission

May 1, 2022
Author notification

July 1, 2022
Final papers with
IEEE copyright forms

The Fourteenth Annual IEEE Energy Conversion Congress and Exposition (ECCE 2022) will be held in Detroit, Michigan, USA,
from October 9 to October 13, 2022. ECCE is a pivotal international event on energy conversion. ECCE 2022 will feature
both industry-driven and application-oriented technical sessions as well as an exposition. The conference will bring together
practicing engineers, researchers and other professionals for interactive and multidisciplinary discussions on the latest
advances in areas related to energy conversion.
Technical papers are solicited on any subject pertaining to the scope of the conference including, but not limited to,
the following major topics:

Energy Conversion Systems & Applications

High power/voltage power conversion Big data and artificial intelligence in energy conversion
High voltage isolation techniques Renewable and alternative energy power
Energy harvesting electronic systems
Energy conversion for information technology Smart grids, microgrids, and utility applications
and communication systems (HVDC, FACTS, and Solid State Transformers)
General Chair Energy efficiency for residential, commercial, Electrical energy storage
Emmanuel Agamloh and industrial applications Wireless power transfer
Baylor University, USA
Component, Converter & Subsystem Technologies
ECCE 2022 Technical Power electronic devices (silicon and wide bandgap) Electromagnetic interference and electromagnetic
Program Co-Chairs and applications compatibility
Navid R. Zargari Passive components and materials Power conversion topologies, modulation, and control
Rockwell Automation, Canada
Power electronic packaging integration Electrical drive systems and topologies and their control
Annette Muetze Reliability, advanced fault protection systems, Rotating/linear electromechanical devices
Graz University of Technology, diagnostics, prognostics, and health management Enabling technologies for Industry 4.0: advanced
Austria
Thermal management and advanced manufacturing, additive manufacturing, digital twins,
Andrea Cavagnino cooling technologies cloud design, big data analytics
Politecnico di Torino, Italy
Mohammad Islam
Halla Mechatronics, USA
Gerry Moschopoulos Digest Submission: Prospective authors are requested to submit a single column, single spaced digest no longer than
University of Western Ontario, five (5) pages summarizing the proposed paper. The digest should include key equations, figures, tables, and references
Canada as appropriate, but no author names or affiliations. Digests not conforming to these requirements will be rejected without
Brandon Grainger review. The digests must clearly state the objectives of the work, its significance in advancing the state of the art, and the
University of Pittsburgh, USA methods and specific results in sufficient detail. All digests will go through a double-blind peer review process to ensure
Contact email: a confidential and fair review. The papers presented at the conference will be included in the IEEE Xplore Digital Library.
ecce2022tpc@gmail.com Please refer to the conference website for a detailed list of technical topics and the digest submission method.

www.ieee-ecce.org/2022 Detroit, Michigan, USA – October 9–13, 2022

Digital Object Identifier 10.1109/MELE.2021.3118233

 IEEE VOL. 9, NO. 4 DECEMBER 2021 ISSN 2325-5897

WWW.IEEE-PES.ORG/

MAGAZINE

F E AT U R E S

8 The New Kid

on the Block
47 Utility-Scale Shared
Energy Storage
Battery energy storage Business models for
systems and hybrid plants. utility-scale shared energy
John Paul “JP” Skeath, Ryan Quint, storage systems and
Julia Matevosyan, Rachana Vidhi, customer participation.
Gary Custer, Spencer Nervig, Mohammed Ben-Idris,
Eric Miller, Paul Spitsen, Michael Brown, Matthew Egan,
Venkat Konala, Songzhe Zhu, Zhenyu Huang, and Joydeep Mitra
and Andrew Isaacs
55 Evolving Toward
19 Grid-Scale Energy
Storage Systems
a Scalable Hyperloop
Technology
Cover image: Addressing the important aspects of
large-scale energy storage deployment.
Ensuring safety. Vacuum transport as
©SHUTTERSTOCK.COM/CYBRAIN
Babu R. Chalamala, David Rosewater, a clean alternative to
Yuliya Preger, Reed Wittman, short-haul flights.
Joshua Lamb, and Akira Kashiwakura Jonas Kristiansen Nøland
MISSION STATEMENT: IEEE Electrification Maga-

67
zine is dedicated to disseminating information on
29 The Codes and
Standards Facilitating
Evacuated-Tube, High-
Speed, Autonomous
all matters related to microgrids onboard electric
vehicles, ships, trains, planes, and off-grid applica-
tions, including electrification of remote communi-
the Design and Maglev (Hyperloop) ties and decarbonization strategies. Microgrids
Adoption of Energy Transport System for refer to electric networks in a car, a ship, a plane,
or an electric train, which have a limited number of
Storage for Power Long-Distance Travel sources and multiple types of loads as well as off-
System Applications An overview. grid applications that include small scale multi-car-
Federico Lluesma, Antonio Arguedas, rier energy systems supplying electricity and heat
Keeping pace with evolving in areas away from high voltage power networks.
safety codes and standards. Sergio Hoyas, Alberto Sánchez, Feature articles focus on advanced concepts, tech-
Jim McDowall, Matthew Paiss, and Juan Vicén nologies, and practices associated with all aspects
of electrification in the transportation and off-grid
and Michael Ropp sectors from a technical perspective in synergy
with nontechnical areas such as business, environ-
38 Emerging Best Practices
for Modeling Energy
mental, social and policy topics.

Storage in Integrated IEEE Electrification Magazine (ISSN 2325-5897) (IEMECM)

is published quarterly by the Institute of Electrical and
Resource Plans Electronics Engineers, Inc. Headquarters: 3 Park Avenue,
17th Floor, New York, NY 10016-5997 USA. Responsibility
An overview and a comparison. for the contents rests upon the authors and not upon the
Jeremy B. Twitchell IEEE, the Society, or its members. IEEE Operations Center
(for orders, subscriptions, address changes): 445 Hoes
and Alan S. Cooke Lane, Piscataway, NJ 08854 USA. Telephone: +1 732 981
0060, +1 800 678 4333. Individual copies: IEEE members
US$20.00 (first copy only), nonmembers US$140.00 per
copy. Subscription Rates: Society members included with
membership dues. Subscription rates available upon
request. Copyright and reprint permissions: Abstracting is
D E PA R T M E N T S & C O L U M N S permitted with credit to the source. Libraries are permitted
to photocopy beyond the limits of U.S. Copyright law for the
private use of patrons 1) those post-1977 articles that
carry a code at the bottom of the first page, provided the
2 ABOUT THIS ISSUE per-copy fee indicated in the code is paid through the
Copyright Clearance Center, 222 Rosewood Drive, Danvers,
4 TECHNOLOGY LEADERS MA 01923 USA; 2) pre-1978 articles without fee. For other
copying, reprint, or republication permission, write
74 DATES AHEAD Copyrights and Permissions Department, IEEE Operations
Center, 445 Hoes Lane, Piscataway, NJ 08854 USA.
76 NEWSFEED Copyright © 2021 by the Institute of Electrical and
Electronics Engineers, Inc. All rights reserved. Postmaster:
84 VIEWPOINT Send address changes to IEEE Electrification Magazine, IEEE
Operations Center, 445 Hoes Lane, Piscataway, NJ 08854
USA. Canadian GST #125634188 PRINTED IN U.S.A.

ITEC+ will be held 15–17 June 2022 at the Westin

Anaheim Resort in California.

Digital Object Identifier 10.1109/MELE.2021.3115536 IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 1

 ABOUT THIS ISSUE EDITORIAL BOARD Silva Hiti
Editor-in-Chief Rivian
Lingling Fan California, USA
University of South Florida silvahiti@rivian.com
Florida, USA Eduard Muljadi
linglingfan@usf.edu Auburn University
Alabama, USA

Energy Storage—
Associate Editors mze0018@auburn.edu
Electric Trains
Tamas Ruzsanyi Mahesh Krishnamurthy
Ganz-Skoda Illinois Institute

The Buffer We Have

Hungary of Technology
tamas.ruzsanyi@ieee.org Illinois, USA
kmahesh@iit.edu
Eduardo Pilo de la Fuente

Been Looking For

University Francisco Vitoria Electric Planes
Spain Syed A. Hossain
eduardo.pilo@ufv.es GE Aviation
Ohio, USA
José Conrado Martínez syed.hossain@ge.com
Administrador de
Infraestructuras Kaushik Rajashekara
Ferroviarias University of Houston
By Zhenyu Huang and Mohammed Ben-Idris Spain Texas, USA
jcmartinez@adif.es ksraja@central.uh.edu
Babak Nahid-Mobarakeh
Microgrid McMaster University
Suryanarayana Doolla Canada
Indian Institute of Babak.nahid@ieee.org
Technology Bombay
India Bulent Sarlioglu
OR MORE THAN A CENTURY, THE POWER SYSTEM HAS University of

F
suryad@iitb.ac.in
been an amazing balancing act—how much electricity Wisconsin-Madison
Mohammad Wisconsin, USA
is generated has to be exactly how much is consumed. Shahidehpour bulent@engr.wisc.edu
This is quite a unique situation. Most other complex systems— Illinois Institute
of Technology IEEE PERIO D IC ALS
natural or man-made—have buffers built in: rivers can have res- Illinois, USA MAGAZINES
ervoirs; supply chain systems have warehouses; and gas ms@iit.edu DEPARTMEN T
445 Hoes Lane,
systems have tanks. These buffers decouple the flows and make Steve Pullins Piscataway, NJ 08854 USA
system operation more flexible, and fluctuations can be con- AlphaStruxure
Craig Causer
Massachusetts, USA
tained in local areas. Without these buffers, any imbalance in Managing Editor
steve.pullins@se.com
the power system can cause system-wide operation issues and Geri Krolin-Taylor
Antonello Monti Senior Managing Editor
potentially lead to outages and/or equipment damages. This RWTH Aachen
becomes even a more challenging situation because of larger Germany Janet Dudar
amonti@eonerc Senior Art Director
fluctuations due to the ever-increasing renewable penetration. If
.rwth-aachen.de Gail A. Schnitzer
the power system has large buffers, many of the operational
Associate Art Director
challenges would no longer exist, complexity would be reduced, Electric Ships
Marta Molinas Theresa L. Smith
and flexibility would increase. Energy storage is such a buffer for Norwegian University Production Coordinator
which power engineers have been looking. However, to truly of Science and Felicia Spagnoli
Technology
have the benefit from energy storage, it has to be deployed at Advertising Production
Norway Manager
scale, and it needs to be long duration for mitigating the season- marta.molinas@ntnu.no
al and diurnal nature of renewable generation. Large-scale ener- Peter M. Tuohy
Herb Ginn Production Director
gy storage deployment needs safety assurance, codes and University of
South Carolina Kevin Lisankie
standards, and methods and tools for system-level planning, Editorial Services Director
South Carolina, USA
control, and operation. ginnhl@cec.sc.edu Dawn M. Melley
In this special issue of IEEE Electrification Magazine, we present Robert Cuzner
Senior Director,
Publishing Operations
five feature articles and two columns that address several impor- University of
tant aspects of large-scale energy storage deployment. This issue Wisconsin-Milwaukee
ADVERTISING SALES
Wisconsin, USA
also includes two interesting feature articles on the hyperloop Beverly Grimme
Cuzner@uwm.edu
Account Executive
technology that has great potential to reduce carbon emissions in Naylor Association Solutions
Electric Vehicles
the transportation sector but adds additional requirements on the Khwaja Rahman
Direct: +1 352 333 3367
Cell: +1 904 881 0862
electricity sector, which energy storage can play an important role Rivian
Michigan, USA Fax: +1 703 790 9199
to support. krahman@rivian.com bgrimme@naylor.com
The issue starts with the “Technology Leaders” column, “The
Future of Electricity Storage: From Hours to Days,” by Paul

Digital Object Identifier 10.1109/MELE.2021.3115538

Date of current version: 1 December 2021 Digital Object Identifier 10.1109/MELE.2021.3115537

2 IE E E E l e c t r i f i c ati o n M agaz ine / DECEMBER 2021

Albertus. The author articulates the
importance of long duration energy
storage for time shift of renewable
generation to mitigate seasonal and
diurnal issues. He also indicates that
energy storage with 12-h duration is
necessary for a power system with a
50% or higher penetration of wind
and solar generation.
In the first feature article, “The
New Kid on the Block,” Skeath et al.,
discuss the applications and benefits
from battery storage and hybrid
power plants. It offers a full view of
design, planning, and operation con-
siderations at the component, plant,
and system levels of battery storage
and hybrid power plants. The article
emphasizes that battery storage and
hybrid plants bring more than one
application to the system and such
multiple applications should be fully
utilized to support the bulk power
system reliability.
In the second feature article, “Grid-
Scale Energy Storage Systems,” Chala-
mala et al., review the safety hazards
large-scale energy storage of various
kinds—lithium-ion battery systems,
aqueous battery systems, pumped
hydro storage, and fly wheels—would
bring to the power system, the envi-
ronment, and people, with a particu-
lar emphasis on system-level safety
issues. For example, even though the
failure rate of a single battery cell is
very low (< one in 1 million), there are
100,000 cells in a large-scale MWh
battery system. The system-level fail-
ure rate is much higher.
In the third feature article, “The
Codes and Standards Facilitating the
Design and Adoption of Energy Stor-
age for Power System Applications,”
McDowall et al., discuss recent devel-
opment of codes and standards for
battery systems, concerning safety,
integration, and operation. The codes
and standards include IEEE P2800,
IEEE 1547, IEEE 929, IEEE 1679, UL
1973, UL 1991, UL 1998, UL 9540, UL
1741, and the International Fire Code
and the National Fire Protection
Association fire codes, across compo-
nent and system levels.
In the fourth feature article,
“Emerging Best Practices for Model-
ing Energy Storage in Integrated
Resource Plans,” Jeremy Twitchell and
Alan Cooke examine an important
step for large-scale energy storage
deployment, i.e., integrated resource
planning (IRP). With their uniquely
flexible and scalable nature, energy
storage technologies can provide a
wide variety of grid services for the
future power system. Such grid ser-
vices need to be properly represented
in the IRP process. This article identi-
fies specific practices at leading utili-
ties for accurately evaluating the
costs and benefits of energy storage
in the IRP process. By identifying
these practices, this article can
inform utilities, regulators, and other
IRP stakeholders in improving the
representation of energy storage in
IRP processes.
In the fifth feature article, “Utility-
Scale Shared Energy Storage,” Ben-Id-
ris et al. discuss the implementation
and application of utility-scale
shared energy storage (USES) sys-
tems and challenges related to build-
ing business models that manage
interactions between utilities and
customers. The article also reviews
potential applications and grid ser-
vices as well as equity and reliability
USES can offer; technical challenges
and requirements to implement
these services are discussed. Survey
data for customer willingness to
lease virtual blocks of USES systems
are also presented.
The sixth feature article focuses
on the hyperloop technology. “Evolv-
ing Toward a Scalable Hyperloop
Technology” is authored by Jonas
Kristiansen Nøland and provides a
different but relevant view on ener-
gy storage in future hyperloop appli-
cations. Hyperloop provides an
alternative approach to conventional
transportation with reduced carbon
emissions, among many other
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1
benefits. The pulsating nature of the
hyperloop operation creates a highly
stressful load profile. The onboard
energy storage system can be used to
mitigate this pulsating issue in a
cost-effective manner.
The seventh feature article is our
second on hyperloop technology.
“Evacuated-Tube, High-Speed, Auton-
omous Maglev (Hyperloop) Transport
System for Long-Distance Travel” by
Lluesma et al., continues the discus-
sion of the hyperloop technology by
comparing multiple implementa-
tions for long-distance travel in
terms of their power and pressure
requirements. Its comparison with
conventional aircrafts clearly indi-
cates hyperloop’s advantages on
global efficiency by always operating
within designed conditions. This arti-
cle also points out the carbon reduc-
tion benefits by hyperloop.
This issue wraps up with the
“Viewpoint” column, where Eric
Hsieh presents “Scaling up to Long-
Duration Storage.” The author re-
views the historical perspective of
energy storage development from
the first MW-scale pumped hydro
storage in late 1920s to today’s more
than 20-GW storage in operation.
Such development is compared with
that of parallel technologies such as
LED lighting, modern computing, and
power generation, to gain insights of
accelerating diverse and disruptive
technologies. The U.S. Department of
Energy’s Long Duration Energy Stor-
age Earthshot uses the metric kWh-
cycle and sets a target of 5 cents per
kWh-cycle Levelized Cost of Storage
to drive energy storage development.
Enjoy the Issue
We hope you enjoy the diverse and
comprehensive views in this issue.
If you would like to submit an arti-
cle or would like a specific topic to
be addressed in future issues,
please contact the editorial board at
electrification@ieee.org.

3
 TECHNOLOGY LEADERS
The Future of Electricity Storage:
From Hours to Days
By Paul Albertus
L E C T R I C I T Y S YS T E M S
E have long relied on exten-
sive amounts of fuel stor-
age to avoid supply disruptions and
ensure that electricity can be deliv-
ered as needed. On-site storage of
nuclear fuel and coal, as well as
pipeline and underground storage of
natural gas, are typically able to pro-
vide conventional generators with
weeks to years of secure fuel supply,
and lack of access to fuel by large
scale-generators is rare. However, as
electricity supply systems shift to
increased use of the variable renew-
able energy sources (RESs) wind and
solar, due to favorable cost and emis-
sions attributes, there are funda-
mentally new challenges for grid
planners and operators as the
annual share of electricity from
these sources reaches 50% and
higher. These trials include faster
ramp rates for conventional assets,
a reduction in the inertia of rotating
generators, and spatial changes in
generation assets (and hence altera-
tions in transmission and distribu-
tion networks) as well as market
challenges around adding a large
fraction of generation assets with
no fuel costs and a limited ability
for scheduling.
To address these, there are sever-
al technical options to expand the
Digital Object Identifier 10.1109/MELE.2021.3115539
Date of current version: 1 December 2021
4 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
use of intermittent wind and
solar while ensuring a stable sup-
ply, including
1) expanding the transmission net-
work to enable greater spatial
averaging of variable generators
2) expanding load flexibility through
demand response and other pro-
grams for industrial and residen-
tial customers
3) ramping conventional genera-
tors, such as natural gas peakers,
potentially including carbon cap-
ture and sequestration to limit
greenhouse gas emissions
4) curtailing intermittent resources
beyond those needed at the time
of production
5) storing electrical energy generat-
ed by wind and solar.
While the first four options are
being pursued, they have significant
limitations. For example,
1) expanding the transmission net-
work is slow because of permit-
ting challenges associated with
public resistance
2) expanding load flexibility has dif-
ficulties achieving significant
load reductions because of cus-
tomer hesitancy about the
programs
3) ramping conventional generators
still results in emissions and if
combined with carbon capture is
quite expensive
4) curtailment increases costs and
will eventually require the building
of less favorable projects as the
best sites are taken.
Hence, there are significant
opportunities in the fifth option, elec-
trical energy storage, due to its
potential to provide truly large-scale
(i.e., hundreds of gigawatts) electrici-
ty time shifts without emissions or
significant permitting problems.
The role of energy storage has
long been recognized, as evidenced
by the presence of pumped-storage
hydro (PSH) facilities, which in the
United States have a power capacity
of around 23 GW, with an average
duration of roughly 10 h, for a total of
approximately 230 GWh. PSH is by
far the largest form of electrically
charged storage in the world today.
However, there are only a few new
large-scale facilities being construct-
ed because of environmental and
other constraints. While a number of
thermal storage facilities integrated
with concentrating solar power
plants have been and will be built,
they are charged with heat rather
than electricity and therefore cannot
be used to store the power from the
large amount of solar photovoltaic
and wind coming online.
Of the new electrically charged
storage facilities being placed in ser-
vice, batteries constitute the largest
fraction by far, specifically lithium-
ion (Li-ion) technology. Li-ion was
developed for portable electronics
and adopted for vehicles, and in the
2325-5897/21©2021IEEE
past decade or so, it has been used for
stationary storage, as well. Automo-
tive applications for Li-ion technology
are driving a truly historic and mas-
sive scale-up in manufacturing, with
projections of annual production of
>1 TWh/year by 2030. For reference, the
average U.S. power demand is about
450 GW, so a single year’s production
of 2 TWh would be enough to time-
shift about 4 h of the entire country’s
average electricity consumption.
Because the technical require-
ments for Li-ion vehicle applications
are not significantly different than
those for the intraday time shift sta-
tionary application, and the cost tar-
gets for automotive applications also
translate to favorable prices for intra-
day time shift stationary applica-
tions, it is clear that there will be
substantial deployment of Li-ion sta-
tionary storage, and, the market has
been growing rapidly in the past
several years. In a case of positive
feedback, recent analysis indicates
that deploying solar and storage
together may increase the market
opportunity for each of those tech-
nologies. Finally, a number of govern-
ments (especially state governments
in the United States) have imple-
mented storage deployment man-
dates, which are serving to spur the
development of stationary storage
industries that will lower the “soft
costs” of project development.
While the challenges presented by
intermittent wind and solar will likely
be economically addressed by Li-ion
(and potentially other storage technol-
ogies, especially for durations in the
range of 6–12 h), there are obviously
time scales of days, weeks, seasons,
and years during which electrical time
shifts may also need to occur. A semi-
quantitative estimate of the maxi-
mum required storage duration versus
the annual fraction of intermittent
electricity on a regional grid [e.g., such
as the California Independent System
Operator (CAISO) and Electric Reliabil-
ity Council of Texas (ERCOT)] is pre-
sented in Figure 1, drawn from a
publication developed from analysis
that led to the Advanced Research Proj-
ects Agency–Energy program Duration
Addition to Electricity Storage (DAYS).
Depending on the amount of curtail-
ment, transmission, and grid flexibility
(i.e., demand response), the maximum
required (where required means the
storage duration to ensure that the
load is met at all points in time) storage
duration remains in the intraday region
(i.e., <12 h) up to about 50–70% of the
annual electricity from wind and solar.
The next region in the graph, “Day/
Week,” spanning from approximately
10 to the low hundreds of hours, could
help a grid reach around 90% of its
annual electricity from wind and solar.

Leaders in grid modernization

have something in common.

They use the RTDS® Simulator

to test their systems.

How do you de-risk control and protection schemes for electrification?

The RTDS® Simulator is the world standard for real time digital simulation,
used worldwide for the closed-loop testing of protection and control systems.

Hardware-in-the-loop testing allows users to de-risk the integration of novel

systems, anticipate and mitigate negative interactions, and optimize
performance by connecting real devices to a simulated network. YOUR WORLD IN REAL TIME.
RTDS.COM
In the hands of innovators, it’s changing the grid as we know it.

CIGRE Electra Ad - Half Page - 2020.indd 1 IEEE Elec trific ation Magazine2021-11-09
/ D EC EM9:30:12
BE R AM
2021 5
 TECHNOLOGY LEADERS

Moving beyond that would require a served primarily by generators but transportation, its use for LDES appli-
seasonal duration. could be met by an LDES technology; cations could leverage those invest-
For reference, the annual fraction of 2) secure self-generation for off-grid ments. Storage in liquids, such as
wind and solar in the annual supply individuals and institutions, such as methanol and ammonia, are also pos-
for CAISO in 2019 was roughly 25%, military bases, in a microgrid; and 3) sible, although, in general, as the
and for ERCOT, it was approximately managing transmission and distribu- number of bonds in the stored com-
21%. Hence, in terms of providing tion congestion (i.e., a wireless alter- pound increases (beyond that of
secure bulk supply, even for systems native). Each of these may serve as a hydrogen), the round-trip efficiency
with relatively high solar (CAISO) and first market for LDES technologies as falls, even as the storage cost declines.
wind (ERCOT), hourly or “intraday” their performance-to-cost ratio Some geology-based approaches are
storage should be sufficient for the matures and they compete for large- under development, with the storage
coming years, although if significant scale grid projects. of water (e.g., in conventional PSH and
increases in wind and solar are Finally, what are potential technol- in an approach such as that pursued
achieved, the 2030s (or sooner if ogies that can serve LDES applica- by Quidnet, with water injected
aggressive emissions reductions are tions? For an LDES technology focused underground) and air (e.g., com-
required) may prove to be a time when on >12 h of duration, key metrics to pressed air with thermal storage). A
truly large-scale hourly or “intraday” look for include the marginal expense number of electrochemical technolo-
storage is required and storage of lon- for adding energy, performance, and gies are also under development for
ger durations also has market penetra- cost for a duty cycle with significant long-duration storage.
tion. While there is not a single shallow cycling and relatively few Li-ion technology can certainly be
definition of “long-duration” electricity deep discharge cycles; safety charac- technically configured to serve
storage (LDES), Figure 1 can provide a teristics; the cost of power (in dollars essentially any duration of interest,
high-level context for the types of stor- per kilowatt); round-trip efficiency; but the cost structure is not favorable
age periods that may be envisioned in cycle life; the need for specialized geo- for periods well over 12 h. Numerous
the coming years and decades. Several logic conditions (e.g., caverns and res- other chemistries have a more prom-
additional publications (see the “For ervoirs); and the price of subscale and ising cost structure, including the
Further Reading” section) also provide full-scale projects, among others. iron-based chemistry under develop-
analysis of long-duration storage. There are several technology classes ment in academic laboratories and at
In addition to the application of of interest. These include power- companies such as ESS and Form
LDES to ensure a secure supply on a to-gas-to-power, with hydrogen or Energy. Zinc (Zn)-based chemistries,
grid with a large fraction of wind and methane as the storage medium. such as Zn/bromine and Zn/air, are
solar, there are a number of addition- Hydrogen, in particular, is receiving potential LDES technologies, as are
al applications. These include 1) resil- major investments, and if it is com- others not described here. Thermal
ience, a market that is presently mercialized in a significant way for energy storage may also be used for
electrical energy storage, although a
challenge for LDES applications
(especially in the tens to hundreds of
lity
Maximum Required Storage Duration

hours) is heat loss, which can be

xibi
1,000 Fle l managed with large containers (to
rid na
(Hours at Rated Power)

G o minimize the surface-area-to-vol-

nd as
,a Se ume ratio), but those require on-site
ion
100 iss construction because they are too
n sm
Tra eek large to ship at their deployed size.
nt, y/W
Da
l me nt,
Innovative approaches associated
10 rtai me and with pumped-heat storage and the
Cu i l
ss rta n,
Le e Cu ssio bility use of very hot storage to improve
i xi
Hourly r
Mo ansm Fle efficiency are currently under devel-
1 Tr Gri d
opment in academia, national labs,
0 20 40 60 80 100 and companies. Most efforts to devel-
op electricity storage based on ther-
Annual Electricity From Wind and Solar on a Regional Grid (%)
mal media are planning for durations
of <20 h. Because a significant appli-
Figure 1. A semiquantitative overview of the maximum duration of electricity storage needed
to ensure demand is met at all times versus the fraction of annual energy from variable genera- cation for LDES technologies is deep
tors, such as wind and solar (Albertus 2020). decarbonization of the electricity

6 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

grid, it is worth mentioning that a
significant competitor is dispatchable
carbon-free power, for example, the
use of natural gas peaker plants with
carbon capture and sequestration.
Because peaker plants often have a
capacity factor of 10% or less, the
capital cost of adding carbon capture
and sequestration technology, espe-
cially to arrest a large fraction of the
carbon dioxide in the output, is a
potential economic challenge.
To conclude, energy storage
deployment (primarily of Li-ion bat-
teries) is rapidly increasing, and
future stationary projects will be
able to benefit from and grow with
the automotive industry’s push
toward vehicle electrification
throughout the 2020s. However, as
electricity systems with >50% of
their annual electricity from wind
and solar are built, large-scale stor-
age at durations of >12 h may be of
increasing importance to ensure a
reliable electricity supply and to
benefit other applications, such as
transmission and distribution sup-
port and resilience.
For Further Reading
A. A. Akhil et al., “DOE/EPRI 2013 elec-
tricity storage handbook in collabora-
tion with NRECA,” Sandia National
Laboratories, Albuquerque, NM, Sandia
Rep. SAND2013-5131, July 2013. [Online].
Available: https://www.energy.gov/
sites/default/files/2013/08/f2/Elec
StorageHndbk2013.pdf
P. Albertus, J. S. Manser, and S. Litzel-
man, “Long-duration electricity storage
applications, economics, and technolo-
gies,” Joule, vol. 4, no. 1, pp. 21–32, Jan.
15, 2020. doi: 10.1016/j.joule.2019.11.009.
P. Denholm, J. Nunemaker, P.
Gagnon, and W. Cole, “The potential for
battery energy storage to provide peak-
ing capacity in the United States,”
Renew. Energy, vol. 151, pp. 1269–1277,
May 2020. doi: 10.1016/j.renene.
2019.11.117.
M. S. Ziegler et al., “Storage require-
ments and costs of shaping renewable
energy toward grid decarbonization,”
Joule, vol. 3, no. 9, pp. 2134–2153, Sept.
18, 2019. doi: 10.1016/j.joule.2019.06.012.
J. A. Dowling et al., “Role of long-
duration energy storage in variable
renewable electricity systems,” Joule,
vol. 4, no. 9, pp. 1907–1928, Sept. 16,
2020. doi: 10.1016/j.joule.2020.07.007.
N. A. Sepulveda, J. D. Jenkins, A.
Edington, D. S. Mallapragada, and R. K.
Lester, “The design space for long-
duration energy storage in decarbon-
ized power systems,” Nature Energy, vol.
6, no. 5, pp. 506–516, 2021. doi: 10.1038/
s41560-021-00796-8.
O. J. Guerra et al., “The value of sea-
sonal energy storage technologies for
the integration of wind and solar
power,” Energy Environ. Sci., vol. 7, no. 13,
pp. 1909–1922, 2020. doi: 0.1039/
D0EE00771D.
Biography
Paul Albertus (albertus@umd.edu) is
the associate director of the Mary-
land Energy Innovation Institute and
an assistant professor of chemical
and biomolecular engineering, both
at the University of Maryland, College
Park, Maryland, 20740, USA.

Know Before You

Submit Nominees

Candidate Profile
Information*
2022 IEEE Including education,
professional history, proposed

PES Awards
citation & notable academic
or industry recognition.

Submit Nominations Today! Significant

Contributions
List of nominee’s compelling
Submissions are due contributions to the power &
energy sector and to the IEEE
31 January 2022 Power & Energy Society (PES).

www.ieee-pes/pesawards
Reference Letters**
Requests will need to be initiated
* Active PES Membership may be required for PES awards eligibility.
by the nominator and the final
** Nominations can be submitted without reference letters.
Final recommendations will need to be received by 31 January 2022
recommendations can be typed
for final review & consideration. or an uploaded PDF letter.
Digital Object Identifier 10.1109/MELE.2021.3126973
 By John Paul “JP” Skeath, Ryan Quint, Julia Matevosyan,
Rachana Vidhi, Gary Custer, Spencer Nervig, Eric Miller,
Paul Spitsen, Venkat Konala, Songzhe Zhu, and Andrew Isaacs

The New Kid

on the Block
Battery energy storage systems and hybrid plants.

NERGY STORAGE PROJECTS, PARTICU-

E
larly battery energy storage systems
(BESSs), have flooded interconnec-
tion queues across North America
seemingly “overnight.” Stand-alone
BESS projects as well as BESSs coupled with
renewable energy generation components—hybrid
plants—are some of the most common resources
being studied for interconnection and will likely
constitute a significant portion of the resource mix
in the future. Lawrence Berkeley National Labora-
tory (LBNL) reports that at the end of 2020, there
was roughly 755 GW of total generation capacity
with 200 GW of energy storage projects in inter-
connection queues.
The rapid increase of BESS and hybrid plants in
the bulk power system (BPS) warrants a look at
where this technology started and how it can sup-
port the BPS moving forward. This article explores
the growing number of BESS and hybrid plants
from different perspectives. BESS and hybrid plant
equipment manufacturers share the latest advanc-
es in equipment capabilities. Plant developers and
designers provide examples of new projects and
engineering considerations. Transmission planners
and operators discuss key focus areas for ensuring
reliable and resilient BPS operation.

The Move to the BPS

Power grids across the United States continue
to migrate toward renewable energy resources.

Digital Object Identifier 10.1109/MELE.2021.3115540

Date of current version: 1 December 2021

8 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021 2325-5897/21©2021IEEE

Renewable generation—predominantly wind and solar
photovoltaic (PV) resources in North America—is inher-
ently variable and uncertain, as the prime mover (e.g.,
wind speed and solar irradiance) is not constant or guar-
anteed. These types of resources have significantly
grown in terms of their share of the generation mix
across North America. As the percentage of renewables
increases, the variability of the overall BPS rises. One
widely known result of high renewable penetration is the
infamous California Independent System Operator
(CAISO) “duck curve” (see Figure 1). Solar PV output
grows in the morning when the load is ramping up and
drops off in the afternoon when the load is still high; this
results in severe net load ramps (the net load is the dif-
ference between the system load and the sum of wind
and solar generation) that have to be managed by other
flexible generating resources. BESSs are poised to be a
solution to support balancing variability and provide
flexibility to the BPS. In the “duck curve” example, BESSs
©SHUTTERSTOCK.COM/FRANK_PETERS
can decrease the need for ramping from traditional rotat-
ing generation. This lessens wear on traditional genera-
tion assets and the risk of import or export violations
between regions.
There is a long history of BESSs in the United States;
however, the industry has experienced a wave of BESS
projects seeking interconnection to the BPS during the
past five years. As expected, with the versatile nature of
BESSs, these projects are being utilized for many applica-
tions. Let us review the following primary applications for
BPS-connected BESSs.
xx Provision of frequency support: This involves the
charging and discharging of active power to regu-
late the grid frequency, providing fast frequency
response (FFR), primary frequency response (PFR),
and secondary frequency response; it is also
called regulation.
■■ The FFR is a subset of frequency support in which
a BESS must rapidly respond, for example, in
~250 ms, to frequency deviations.
xx Provision of voltage and reactive power support: This
relates to injecting or absorbing reactive power to
regulate the grid voltage. Projects with reactive
power as a main application are sized much differ-
ently, as they must consider reactive power capa-
bility first.
xx Energy arbitrage: This is the process of discharging
active power during peak consumption and recharg-
ing during times of abundant generation.
xx Participation in capacity market: This application
offers BESSs to a capacity market. Projects vary based
on market specifics, but a general “rule of thumb” is
4-h BESSs.
xx Firming renewable output: This concerns discharging
active power when the output from solar is low and
charging when solar output is abundant. This is typi-
cally done when there is cloud cover.
xx Providing or enabling black-start capability: This
involves using a BESS as a black-start unit in small
island and microgrid applications as well as to start
larger synchronous generators.
While each of these applications has unique behav-
ior, it is common for BESS projects to include multiple
functions. Applications may dictate how transmission
planners model and study BESS projects. Alterations
after construction may have negative impacts on bat-
tery chemistry and lifecycles and trigger a restudy, so
careful consideration of a resource’s uses should occur
prior to selecting a technology. Now, let us say that a
developer or owner needs to come back and change
something from the initial selection. Augmentation is
the ability to combine BESS technologies. BESS plant
age and expansion drives the need for it. Battery tech-
nology has changed and will continue to, and the same
is true for inverter technology. DC–dc converters facili-
tate connecting battery systems of differing dc
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 9
 28,000 Actual 3-h
Ramp of
26,000
17,259 MW
On 28 February 2021
24,000

22,000
2012
20,000 (Actual)
2013 (Actual)
(MW)

18,000

16,000 2014
2015
2016 Net Load of
14,000 2018 2017
3,614 MW
2019 On 13 March 2021
12,000 2020
Over-Generation
10,000 Risk

0
12 a.m. 3 a.m. 6 a.m. 9 a.m. 12 p.m. 3 p.m. 6 p.m. 9 p.m.
Hour

Figure 1. A typical spring day for the CAISO. (Source: CAISO; used with permission.)

voltages to one inverter by matching the converter out- applications, and safety considerations, among others.
put voltage to an existing battery system. Figure 2 When developers are looking to obtain the various
illustrates BESS augmentation. components of a BESS, they can buy the major equip-
ment separately or choose a turnkey solution from an
Selecting and Determining a Technology: engineering procurement company. Whoever is
Preventing Choice Overload responsible for hardware integration (a visual process
Many factors need to be evaluated when selecting and is presented in Figure 3) must focus on the following
quantifying inverters and battery products to meet major components:
project requirements. These include dc voltage range x  Battery supplier and chemistry: Most BESSs utilize lithi-
compatibility for active and reactive power provision, um-ion (Li-ion) battery cells packaged into modules.
applicable interconnection requirements, intended The Li-ion chemistry and the form factor (pouch, pris-
matic, or cylindrical) of the cells are
generally determined by original
BESS Inverter equipment manufacturers (OEMs).
Grid
Integrators will generally work with
OEMs to identify the number of
modules (“battery racks”) that will
dcc need to be stacked in series to
DC–dc
C–dc dc
dc achieve the voltage range needed
erter
Converter
for the selected inverters.
x Inverter: Battery inverters have
had an increasing operating
voltage range during the past
several years, resulting from a
desire to add more modules
Augmented Existing behind each one.
Batteries Batteries x Battery containers: While some
battery OEMs provide a contain-
erized solution, many provide
only battery racks, and it is up to
an integrator to find manufac-
Figure 2. The augmentation of bulk connected BESSs. (Source: SMA America; used with turers that can design and build
permission.) containers for the batteries.

10 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

 Container design typically includes a suitable cooling/ may reduce the inverter active power, while a higher one
heating system, a fire suppression system (if needed), may lessen the inverter overexcited reactive power capa-
and safety controls for access doors, temperature sen- bility. If evaluations are done well using all project
sors, gas detection/ventilation, and so on. assumptions, they should result in a reliable value for the
xx Cooling system: Batteries generate a significant number of inverters required for a project.
amount of heat while cycling, which must be removed Next comes consideration of the intended application
quickly and efficiently. Sizing a heating, ventilation, and safety considerations of a BESS design. Adding racks
and air conditioning (HVAC) unit or other system for to a system increases the batteries’ short circuit potential
battery chemistry and application requirements is one under a fault. BESS projects are classified as either “power”
of the most important BESS safety considerations. or “energy” systems based on their use case. BESSs catego-
These units can be mounted on the side of a battery rized as “power systems” have a shorter energy duration
container (see Figure 4). and are used as, for example, frequency support. With a
xx BESS controls: An energy storage system is not com- shorter discharge time, fewer battery racks need to be con-
plete without a BESS controller. This can be a single nected to a single inverter. Such a system would have a
unit or multiple ones working together. It is vital that short circuit capacity of <150 kA in its batteries. Fuse man-
a controller has all required inputs from the individual ufacturers have been working to keep up with new
components in a system, such as the battery manage- demands from the energy storage market, and there are
ment system (BMS), inverter, and any other safety- many options to protect dc systems in this range.
related devices that must be regulated during a fault. Energy systems, on the other hand, often discharge
The controller will also manage the system power power for long durations. This results in a battery short cir-
flow to meet use case requirements. cuit potential of >300 kA. It is common to find inverter
When pairing an inverter with a battery, a simple first solutions with split bus inputs that enable proper fuse
step is to compare the inverter dc voltage operating range coordination and system protection by using balance-of-
to that of the battery. It is very important to understand system equipment. Always confirm with a manufacturer
the battery’s actual operational voltage range, as it will that a battery is suitable for an application and its available
reflect how the battery will operate in the field. For exam- kiloampere rating. Inverter manufacturers should have
ple, the operational minimum dc voltage is often much guidance based on full short circuit testing. Results from
higher than at a 0% state of charge
(SOC). The output ac voltage and
internal component design can
Battery Cell Battery Module Battery Racks Battery Container
also impact the operational dc volt-
age range.
The next step is to compare
applicable interconnection and
design requirements against the
capability of the inverter-plus-
~7
7 MWh Energy/
battery system. Each inverter
40-ft Container
manufacturer should have a tool
to calculate an inverter’s active
Figure 3. The integration of hardware in battery containers for BESSs. (Source: NextEra; used
power–reactive power (P–Q) capa- with permission.)
bility, which is based on design
assumptions that include a battery’s maximum and mini-
mum dc voltage. It is important to understand how the
battery’s minimum and maximum dc voltage may affect
the inverter power output, especially at the project bound-
ary operating conditions. For example, the minimum bat-
tery dc voltage has a direct relationship to an inverter’s
reactive power capability, specifically, the overexcited reac-
tive power.
It is also important to review a P–Q diagram of the
HVAC Power Controls
inverter when the battery is at its maximum operating
Programmed to Match Managed By Inverters
voltage to account for any derating based on the inverter Charge and Discharge
design. Another design review should look at P–Q dia- Heat Load
grams to evaluate the reactive capability based on
required grid voltage tolerances and temperatures, as Figure 4. The HVAC and power controls needed for bulk BESSs.
those influence the diagrams’ shape. A lower ac voltage (Source: NextEra; used with permission.)

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 11

 these evaluations provide assurance that long-duration
systems have proper fuse coordination and the ability to
withstand design faults. Once the preceding are obtained,
a developer can confidently begin building an online sam-
ple of how the controllers will interface (Figure 5).
The following discussion of inverter-based resources
applies to BESSs, as well. Inverter-based resources are
available as grid-following (GFL) and grid-forming (GFM)
devices. GFL inverters are controlled as current sources,
and GFM variants are managed as voltage sources. Histori-
cally, inverters have been introduced to the BPS in GFL
mode. They operate as a current source with synchroniza-
tion to the grid, enabling them to achieve requested set
points faster, as they directly control the current. GFL
inverters that provide grid services, such as frequency
support, voltage support, and solar plant ramp rate con-
trol, are sometimes called grid supporting.
GFM inverters have risen from microgrid applications
as well as the black-starting of critical grid infrastructure,
and they are currently being considered for providing iner-
tia, improving the voltage stability of weak grids, and
black-start applications. In GFM mode, inverters can con-
trol active and reactive power by directly adjusting the
output voltage, similar to synchronous generators. In the
future, the industry will look to have GFM inverter capabil-
ity lead the transition to an inverter-based grid.
Hybrid Plant Design Considerations
One of the most common forms of hybrid plants in the
BPS is a solar and BESS facility. The surge of solar-plus-
storage stems from BESSs being eligible for investment
tax credits if >75% of their charging energy comes from
solar during a recapture period of ~5 years. Hybrid plants
can be built using an ac-coupled or a dc-coupled configu-
ration. Each of these can make a portion of the solar ener-
gy dispatchable.
AC-Coupled Systems
In this configuration, BESSs and solar are connected on
the low-voltage side of the generator step-up transformer.
BESSs are typically located close to substations to reduce
line losses. They are integrated in nearly the same way as
stand-alone projects and can function independently if
needed. Augmenting new batteries is simpler and the pre-
ferred option if a solar project is being retrofitted with
BESSs. Another advantage of this design is the metering
convenience for the separate market participation of the
two resources. However, the major disadvantage is the
loss of energy on the dc side of the solar inverters, due to
the dc overbuild of a solar array. An example of this con-
figuration is in Figure 6.
DC-Coupled Systems
This configuration utilizes dc–dc converters to couple bat-
tery containers to solar arrays on the inverters’ dc bus.
BESSs have access to the dc-clipped energy that can be
12 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
captured in the batteries, which otherwise would be lost.
Close coupling (as shown in Figure 7) of solar and batteries
results in higher efficiency and increased generation com-
pared to ac-coupled systems. Augmentation and meter-
ing, however, are more challenging, and developers need
to plan for both.
Navigating the Interconnection Process
For a new developer, and sometimes even an experienced
one, navigating the interconnection process is like step-
ping into a maze. You know where the exit is, but uncer-
tainties and variations along the way may lead to a dead
end. Every year, utilities receive hundreds of new intercon-
nection requests. Most are for renewables, BESSs, and
hybrid generating facilities, and only a small number suc-
cessfully complete the process and get developed. Inter-
connection is initiated by developers. They choose the
location, technology and size of generating facilities.
Transmission planners respond to requests by performing
planning studies. Developers modify their requests (and
even cancel projects) based on the findings.
Initiating an Interconnection Request
Some planning authorities provide information to guide
developers toward areas where there is abundant trans-
mission capacity and help them avoid locations where
expensive upgrades are needed. However, transmission
capacity does not always align with the abundance of
renewable resources. Developers often find themselves in
wind- and solar-rich areas that have limited transmission.
Alth­ough adding BESSs could improve the situation,
expensive upgrades are often necessary even for hybrid
interconnection requests. Decisions about who utilizes
limited capacity and gets to the finish line are outcomes
of joint efforts between developers and planning authori-
ties. Some utilities have proactive planning processes that
allocate available capacity to the best standing request.
Others rely on passive procedures that wait for inviable
requests to be withdrawn.
Stand-alone storage in high load and congestion areas
can provide additional benefits to the grid through various
services, such as peak load reduction, voltage support, grid
strengthening, and frequency response. Unlike frequency
regulation, these do not result in battery charging during
peak loads. It should also be noted that in high-density
areas, transmission and distribution upgrades may not be
feasible, and appropriate storage use may be the most
cost-effective solution for rate payers. A collaborative
approach between developers and transmission planning
authorities to determine use cases is needed to ensure
that infeasible and unnecessarily expensive upgrades are
not imposed.
Interconnection Study Process
Most utilities in the United States follow Federal Energy
Regulatory Commission (FERC) pro forma large-generation
 PV-Plus-Storage
Plant (ac Coupled)

Plant SCADA
Power
Operations Dispatch
Communication Center Controller

SCADA and Plant Grid Generator

Database Controller Interface Operator
Historian
PCS Skid
PV Array
Network
Switch
DS Controller
DC/ac Transformer
Inverter
DS PV
Meter

BESS
BESS Skid Controller
Substation
Network POI
BMS Switch Meter
Fire PCS Skid
Protection
System

HVAC
Controller
Gas
Detection DC/ac Transformer
and Enclosure Inverter
Ventilation BESS
PLC Meter

Figure 5. The control of an ac-coupled BESS. SCADA: supervisory control and data acquisition; PCS: power conditioning system; POI: point of interconnection. (Source: Urban Grid; used with permission.)

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1

13
 interconnection procedures. The process consists of feasi- base interconnection studies on queue priority; i.e., all ear-
bility, system impact, and facilities studies. Feasibility lier interconnection requests are assumed in a study.
studies are optional and provide a high-level, early evalu- When an earlier request drops out, a restudy may be
ation of a proposed interconnection. System impact stud- needed. If a queue grows very large and has projects rou-
ies evaluate the effect of a proposed interconnection on tinely being withdrawn, the process can result in endless
the reliability of the transmission system. Finally, facilities restudies and cost vulnerability.
studies specify and estimate the cost of equipment, engi- Understanding how important timeliness and cost
neering, procurement, and construction to implement a certainty are to developers, some utilities have adopted
system impact evaluation. Their end product is a list of procedures. One example is the California ISO’s intercon-
equipment to interconnect generating facilities, provide a nection process, as detailed in Figure 8. There are two
requested interconnection service, and maintain system significant features of the California ISO’s process, as in
reliability. Equipment is categorized as interconnection the following:
facilities and network upgrades. Interconnection facility xx Phase 1 and 2 interconnection studies set the maxi-
costs and scopes are standardized within each utility. mum cost responsibility and maximum cost exposure
With some basic information about the point of inter- for network upgrades. The maximum exposure is esti-
connection, a utility can provide a customer with a good mated by the ISO, based on planning studies, and the
estimate of the direct costs for the facilities. The expense risk of exceeding these costs is born by transmission
to upgrade a surrounding network depends on the trans- owners. Developers are protected by the maximum
mission capacity and status of prior interconnection price exposure and will never be assigned a network
requests, and these can be highly variable. Many regions upgrade cost exceeding it.
x  A restudy is annually performed

for all active interconnection
Babcock Ranch Storage: 10 MW/4 h requests. Impacted customers
receive updates about the scope
and cost of facilities and upgrades.
In the Electric Reliability Council
of Texas (ERCOT) footprint, develop-
ers are not responsible for the costs
Tie-in Location to
Existing Solar of network upgrades beyond their
Storage Equipment Area
point of interconnection. In the full
interconnection study phase, trans-
mission limitations associated with
a project are pointed out to a devel-
oper. Transmission reinforcements
that are necessary to remove these
restrictions will be subject to a
regional planning review. A trans-
Figure 6. An ac-coupled hybrid plant. (Source: NextEra; used with permission.) mission reinforcement project
must be needed for system reliabil-
ity or fulfill socioeconomic benefit
criteria. Thus, project developers
Storage Equipment Adjacent are exposed to uncertainty associ-
to Solar Inverters ated with available transmission
capacity. If a project is connected to
a region with insufficient capacity,
it may have its output curtailed.
Even if there are no transmission
limitations during the interconnec-
tion process, future projects in the
same area may lead to power out-
put restrictions.
Regardless of the process, all
Tie-in Location at
interconnection studies include a
Multiple Solar Inverters power flow analysis (which evalu-
ates fundamental physical limits in
Figure 7. A dc-coupled hybrid plant. (Source: NextEra; used with permission.) transmission elements), stability

14 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

analysis (which examines the dynamic response of a sys-

Figure 8. The California ISO generation interconnection procedures. TPD: transmission plan deliverability; IR: initial request; IFS: interconnection financial security. (Source: CAISO; used with permission.)
tem to critical events), and short circuit analysis (which
ensures that the fault-arresting capacity of substation

Commercial
Operation
equipment, such as protective breakers, is not exceeded).
Modeling inverter-based resources has imposed new chal-
lenges to transmission planners and increased the scope
and complexity of these studies. This is especially true for
BESS and hybrid plants, due to more permutations of pos-
sible operating conditions. Planners need to study differ-

Initial Synch
ent system load conditions (peak, off-peak, and shoulder
peak) with different BESS dispatch assumptions, such as
discharging and charging. A BESS can go through the full
cycle of maximum charge to maximum discharge during
the peak load hour. Therefore, examining maximum

Reassessment

Year 3 and on
charging under peak load conditions may be necessary.

March~July
Perhaps more importantly, dispatch assumptions for

Annual
all resources should be consistent with load conditions.
For example, when a battery is charging under peak load,
solar resources are near their maximum output. There-
fore, battery charging is less likely to cause transmission
system overloads for hybrid and BESS plants next to solar

IFS Posting
Second

Year 3
ones. The growing number of study scenarios has put

May
many planning entities under pressure and adds to the
schedule and cost risk of interconnection projects.
Some interconnections pose technical challenges that
cannot be evaluated using standard power flow and TPD Allocation

January~March
TPD Allocation
dynamic tools. Increasingly, BESSs and other inverter- Choices
Interconnection Customer Decision

based resources connected to weak network areas and in

Year 3
particularly complex regions may be subject to additional
studies employing more detailed tools, termed electromag-

Study Process
netic transient (EMT) tools. These require additional time
and expertise from planning teams, and extra studies can

May~November
Phase 2 Study
further complicate and delay interconnection processes.
Further complications arise from the magnitude of the
data needed to transpose a model into EMT tools. Howev- Year 2
er, with a validated and benchmarked EMT model, a
developer can directly match parameters from the model
to the field, a significant benefit during the interconnec-
tion process.
IFS Posting

Year 2

The difficulty of obtaining good models for inverter-

April
First

based plants, including BESSs, is present during all stages

Megawatts, and so on
Deliverability Option,
Deliverability Status,

of the interconnection process. This is especially trouble-

some during early phases of the interconnection process
when various design decisions are still “in flight” and may
Phase 1 Study

July~January

change. ERCOT has developed modeling and model vali-

Year 1

dation requirements pertaining to each stage of the inter-

connection process, starting after the initial screening
study and extending to the commissioning stage. Models
must be validated by a project developer through a num-
ber of tests, demonstrating comparable performance in
IR Processing

several simulation tools and corroborated by manufactur-

April~June
Year 1

er test results. The modeling and model validation

requirements continuously apply, even after commission-
ing, to make sure models are being updated with the lat-
est changes, such as hardware upgrades and control
setting alterations.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 15

 Modeling control coordination in a
hybrid plant is another challenge. The
industry has extended much effort to
enhance the modeling capability in
power flow and dynamic models. All
major power system simulation soft-
ware platforms have or will release a
power plant controller power flow
model, which helps enforce the plant-
level active power limit and coordi-
nates voltage control among all
components. Entities such as the
Western Electricity Coordinating
Council Model Validation Subcommit-
tee and the North American Electric
Reliability Corporation (NERC) In-
verter-Based Resource Performance
Working Group promote the appro-
priate use of generic models for hybrid plants. However,
there remain challenges in correctly parameterizing the
models and limits within the representations themselves
to correctly describe plant behavior, necessitating more
detailed tools and further studies.
Many interconnection entities are struggling to find
processes and policies for models and studies that main-
tain confidence in future grid reliability while tightening
schedules and minimizing cost risks for projects. As BESS
and other inverter-based resources continue to flood the
system, this pressure is expected to increase. Planning
authorities find themselves at a time when they must rise
to the challenge. Utilities need to invest in their engineer-
ing departments to expand the number of available plan-
ners. Engineering staffs will need to remain proactive in
understanding BESS control logic, model improvements,
and study procedures.
Modifications
In the world of renewable interconnections, very few
requests are completed the way they were initially
planned, and all interconnection study processes include
rules for modifications. Nonmaterial modifications are
allowed throughout the process without impacting a
request’s queue priority. The FERC defines material modifi-
cations as those “that have a material impact on the cost
or timing of any interconnection request with a later
queue priority date.” Each utility has rules and processes
to determine whether a modification is material. Under-
standing and utilizing the modification process is critical
to the success of an interconnection project. In addition
to common modifications, such as inverter and associat-
ed interconnection facility changes, BESS and hybrid
interconnection requests often require changes to the
dc-side battery megawatt-hour volume, charging source,
dc- or ac-coupled configuration, and so on. Most of the
time, these modifications are not material. Early in an
interconnection procedure, when there a lot of
16 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
Many interconnection uncertainties, developers try to con-
figure requests in the most flexible
entities are way for future changes.
Varying interpretations of the
struggling to find FERC material modification rules can
processes and also introduce significant confusion,
delays, and risks into project inter-
policies for models connections, depending on where
the rules are being applied and how
and studies that the study process works. For exam-
maintain confidence ple, a late change by a BESS invert-
er manufacturer may be deemed
in future grid material if a plant is connecting to
a challenging area. These changes
reliability. are common, as inverter technology
is continuously being refined and
OEMs adjust their offerings. However,
if a material change requires a signif-
icant restudy, a project may be forced to restart at the
end of the queue.
In some places, the modification process can be used
by developers to expeditiously add BESSs to existing gen-
erating facilities and active interconnection requests
with an earlier queue date. The FERC mandates that “the
transmission provider must provide a process that allows
an interconnection customer to utilize or transfer sur-
plus interconnection service at an existing point of inter-
connection.” Pairing a BESS with a wind or solar facility
while maintaining the original maximum megawatt
injection at the point of interconnect is the most com-
mon utilization of the surplus interconnection service.
Some utilities, e.g., the California ISO, incorporate the
surplus interconnection service into their modification
processes. As long as a modification is nonmaterial
(which is not always true), a developer can add BESSs
and replace generating facilities with BESSs through the
modification process. This can greatly shorten the time-
line for hybrid project development.
However, as stated, the definition of nonmaterial has not
been generalized across all regions. Outside organized mar-
kets, utilities studying the addition of a storage resource to
a solar or wind facility have varying views about whether or
not a project can be considered a material change. Uniform
and common definitions will help accelerate the adoption
of storage across the country, and interconnection custom-
ers must communicate carefully with regional authorities
to ensure the most efficient path forward.
Operations, the BPS, and
Battery Performance
As an inverter-based resource, BESSs have all the state-
of-the-art capabilities that were developed through the
years for other inverter-based resources, such as wind
and solar. Standard capabilities include voltage support,
frequency support, high and low voltage, frequency ride-
through, and stable operation in weak grid conditions.
While these are normally available
from wind and solar only when they
While BESSs do not
are generating power into the grid, have physical inertia
BESSs can deliver them in charging
and discharging modes and event at like synchronous
zero active power output. However,
the specifics of BESSs produce addi-
machines do, their
tional considerations. controls can be
Systems dominated by synchro-
nous machines exhibit an initial iner- tuned to provide
tial response followed by a slower
turbine governor response or PFR.
suitably fast energy
While BESSs do not have physical injection.
inertia like synchronous machines
do, their controls can be tuned to pro-
vide suitably fast energy injection—
FFR—such that the initial rate of change of frequency and
frequency nadir in low-inertia systems improve. This
aspect of battery storage helps improve system reliability.
Note that voltages should be closely monitored, as high-
speed active power responses can cause high-speed volt-
age fluctuations, especially in weak grid conditions.
Another function of a traditional synchronous genera-
tor is the ability to dampen power system oscillations.
Many synchronous generators are equipped with power
system stabilizers (PSSs) that dampen system oscillations,
typically in the range of 0.2–2 Hz. As these resources
become increasingly scarce, there is a growing need for
oscillation damping support in certain parts of the BPS.
For example, in the Texas Panhandle, which is dominated
by renewable generation, synchronous generators may be
off-line during high renewable output, which could lead to
insufficient damping. Currently, for ERCOT, renewable
generation resources and BESSs are not required to pro-
vide damping support, and synchronous condensers used
to increase system strength in the Panhandle are not
equipped with PSSs. A 2019 ERCOT study identified oscil-
latory behavior of around 1.8 Hz between synchronous
condensers in the region and synchronous generators in
other areas. System planners may identify a need for this
type of control as the penetration of inverter-based
resources continues to increase and define requirements
for inverter-based resources to provide power oscillation
damping. BESSs are well suited to this application.
Care should be taken regarding the SOC since it affects
the ability of a BESS to provide energy and essential reli-
ability services to the grid at any given time. In many
cases, a BESS may have SOC limits that are tighter than
0–100% for the battery life span and other equipment and
performance considerations. These restrictions should be
taken into account while evaluating the ability of a BESS
to provide energy and stability functions.
Grid Forming and Black Starts
BESSs have unique abilities to effectively deploy GFM
technology that helps improve BPS reliability in high
penetrations of inverter-based
resources. Key aspects that enable
this include the availability of an
energy buffer against generation
and load imbalances, low communi-
cation latency between different
controller layers, and robust dc volt-
age that enables the synthesis of an
ac voltage for a wide variety of sys-
tem conditions. In grids where sys-
tem strength and other stability
issues are of concern, BESSs may be
required to support reliable BPS
operation. System planners should
develop interconnection require-
ments and practices as needed to
integrate GFM technology into the planning process,
especially in areas where stability is a worry.
GFM BESSs may have the ability to form and sustain
their own electrical island if they are designed and desig-
nated as part of a black-start cranking path. This may
require new control topologies and modifications to invert-
er settings. Black-start conditions may cause large power
and voltage swings that must be reliably controlled and
withstood. For a BESS to operate as a black-start resource,
energy availability assurance is required. This includes a
specific period for restoration activities and a requirement
that energy be available 24/7. At this time, it is unlikely that
most legacy GFL BESSs can support system restoration
activities as stand-alone resources; however, they may be
used to start subsequent solar, wind, and synchronous
machine plants. GFM capability from BESSs is low-hanging
fruit compared to other inverter-based resources and
should be considered by system planners and developers
to better position themselves for the future.
Resource Adequacy
With BESSs having typical durations of around 4 h, their
role in supporting long-term energy adequacy and assur-
ance is limited. That said, BESSs with 4 h of duration and
longer are participating in capacity markets and remuner-
ated for their contribution to resource adequacy. Storage
for resource adequacy is an active area of research. Find-
ings from the National Renewable Energy Laboratory, Elec-
tric Power Research Institute, and LBNL show that BESS
contributions to resource adequacy is affected by energy
capacity (megawatt-hours) and intended uses. For exam-
ple, BESSs intended solely for frequency response may not
be available to provide energy during peak load conditions.
This is, for example, the case for most BESSs in ERCOT.
The contribution of storage to resource adequacy is
affected by the width of the net load peak and is system
specific. As more storage is added to a system, the width
of the net load peak increases, and contributions from
BESSs with durations shorter than that width will be
reduced, as they need to provide energy through a
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 17
 longer period of time. Adding solar to a system tends to
narrow the peak net load period, potentially increasing
the ability of shorter-duration storage to act as peaking
capacity. Again, the effectiveness depends on regional
demand patterns.
To date, limited information about historic BESS oper-
ation is available. This further restricts the ability to eval-
uate BESS impacts on resource adequacy. Having scarce
historic operation data makes it difficult to develop BESS
dispatch assumptions for transmission planning, opera-
tion planning, and outage coordination studies. Addi-
tionally, there is little information about BESS forced
outage rates, which is used to factor BESS availability in
loss-of-load studies, probabilistic analyses, and other
transmission planning evaluations to assess BPS reliabil-
ity. As BESS penetrations continue to grow, resource ade-
quacy analyses—input data, models, and operational
understanding—need to improve to ensure that they can
be conducted accurately in the future. While the energy
capacity of a BESS project may not be as important sys-
tem stability analyses, this type of information is essen-
tial for resource adequacy assessments.
Focus Areas for Increasing BESS and
Hybrid Plant Deployment
As the system experiences a massive influx of BESS and
hybrid plants in interconnection queues across North
America and around the world, the focus will shift to
ensuring the reliable, resilient, and secure operation of
these generation assets. The growing flexibility of these
resources can equate to increased uncertainty about how
a system will operate in the future, which adds challenges
to planning phases and interconnection processes.
Improvements to the overall interconnection process will
be needed to enable a streamlined timeline and ensure
that adequate models are provided, studies are per-
formed, and reliability issues are addressed. Operators will
need new tools and abilities as conventional resource ade-
quacy methods shift, grid services are provided by these
new assets, and these resources come to be relied on for
energy, capacity, and system restoration. New optimiza-
tion tools, monitoring abilities, operator situational aware-
ness resources, and other innovations will be necessary to
fully enable these assets.
Conclusion
Battery technology is still relatively new to the BPS and
finding a way to “fit in” with the other kids in the neigh-
borhood. BESS projects can provide a variety of support
capabilities to the BPS, and many projects cite more
than one intended application for their facilities. System
operators should fully utilize BESSs to support reliable
BPS operation. Capabilities such as primary and FFR,
GFM, and other functions, which are more readily avail-
able in these new technologies, should be harnessed, as
they are essential for maintaining BPS reliability.
18 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
Equipment manufacturers, developers, owner/operators,
transmission entities, regulatory entities, research insti-
tutes, and academia will need to come together to
advance the technology and its optimal use to ensure
the reliability and resilience of the electrical ecosystem
in the years to come.
For Further Reading
J. Matevosyan, J. MacDowell, N. Miller, B. Badrzadeh, D. Rama-
subramanian, A. Isaacs et al., “A future with inverter-based
resources: Finding strength from traditional weakness,” IEEE
PES Power Energy Mag., vol. 19, no. 6, Nov./Dec. 2021.
“Reliability guideline: Performance, modeling, and simula-
tions of BPS-connected battery energy storage systems and
hybrid power plants,” North American Electric Reliability
Corporation, Atlanta, GA, Mar. 2021. [Online]. Available: https://
www.nerc.com/comm/RSTC_Reliability_Guidelines/Reliability
_Guideline_BESS_Hybrid_Performance_Modeling_Studies
_.pdf
Y. Lin et al., “Research roadmap on grid-forming inverters,”
National Renewable Energy Laboratory, Golden, CO, NREL/
TP-5D00-73476, Nov. 2020. [Online]. Available: https://www
.nrel.gov/docs/fy21osti/73476.pdf
U.S. Department of Energy, “Hybrid energy systems:
Opportunities for coordinated research,” National Renew-
able Energy Laboratory, Golden, CO, DOE/GO-102021-5447,
Apr. 2021. [Online]. Available: https://www.nrel.gov/docs/
fy21osti/77503.pdf
“Report on power plants seeking transmission intercon-
nection,” LBNL DOE/DE-AC02-05CH1123, May 2021. [Online].
Available: https://emp.lbl.gov/publications/queued-character
istics-power-plants
Biographies
John Paul “JP” Skeath (john.skeath@nerc.net) is with
North American Electric Reliability, Atlanta, Georgia, 30326,
USA.
Ryan Quint (ryan.quint@nerc.net) is with North Ameri-
can Electric Reliability, Atlanta, Georgia, 30326, USA.
Julia Matevosyan (julia.matevosyan@ercot.com) is with
the Electric Reliability Council of Texas, Austin, Texas,
78744, USA.
Rachana Vidhi (rachana.vidhi@nexteraenergy.com) is
with NextEra Energy, Juno Beach, Florida, 33408, USA.
Gary Custer (gary.custer@sma-america.com) is with
SMA America, Rocklin, California, 95765, USA.
Spencer Nervig (spencer.nervig@sma-america.com) is
with SMA America, Rocklin, California, 95765, USA.
Eric Miller (eric.miller@ee.doe.gov) is with the U.S.
Department of Energy, Washington, D.C., 20585, USA.
Paul Spitsen (paul.spitsen@ee.doe.gov) is with the U.S.
Department of Energy, Washington, D.C., 20585, USA.
Venkat Konala (venkat.konala@urbangridco.com) is
with Urban Grid, Stevensville, Maryland, 21666, USA.
Songzhe Zhu (szhu@caiso.com) is with the California
Independent System Operator, Folsom, California,
95630, USA.
Andrew Isaacs (ai@electranix.com) is with Electranix,
Winnipeg, Manitoba, R3Y 1P6, Canada.

 Grid-Scale Energy
Storage Systems
By Babu R. Chalamala, David Rosewater, Yuliya Preger,
Reed Wittman, Joshua Lamb, and Akira Kashiwakura
NERGY STORAGE SYSTEMS ARE BECOMING
E
widely deployed throughout the electricity
infrastructure. Large-scale integration of
energy storage systems will become much
more widespread as we begin to integrate
larger amounts of renewables. Furthermore, electrification
Digital Object Identifier 10.1109/MELE.2021.3115541
Date of current version: 1 December 2021
2325-5897/21©2021IEEE
Ensuring safety.
©SHUTTERSTOCK.COM/LAYERACE.COM
of the transportation sector will demand fast charging
infrastructure and energy storage to handle new loads on
distribution systems. However, energy storage systems,
especially battery energy storage systems (BESSs), present
a range of hazards that make engineering safety of large-
scale systems a challenge. This article explores engineer-
ing safety of grid energy storage systems from the
perspective of an asset owner and system operator. We
review the hazards of common lithium-ion and aqueous
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 19
 battery system designs along with the state-of-the-art
hazard mitigation methods. We also summarize the devel-
opment of codes and standards to ensure safety and dis-
cuss best practices for emergency preparedness.
Introduction and Background
Energy storage inherently has hazards associated with
inadvertent release of stored energy. This is true whether
the energy is stored as traditional hydrocarbon fuels, sub-
ject to fires, as pumped hydroelectric storage, subject to
structural failures, or as electrochemical energy storage,
subject to thermal runaway and gas evolution. In the case
of hydrocarbon fuels, we are familiar with the hazards of
fuel fires, and the industry has developed safety systems
to address critical issues through design and globally
adopted safety standards. In the case of pumped hydro,
we reduced the possibility of dam failures through robust
engineering-designs and construction methods. However,
in the case of electrochemical energy storage, the use at
grid-scale is relatively new and safety issues are still
being analyzed.
With a growing need for energy storage in application
markets, such as reliability and capacity firming, the
pace of deployments of energy storage systems for grid
applications is accelerating. Deployment of BESSs as
large as 100 MWh is becoming common and the first
gigawatt hour-sized BESS plants are beginning to be con-
nected to the bulk electric grid, including a recently com-
missioned system at Moss Landing near Monterey,
California. Hybrid systems such as solar + storage are
being built to replace other traditional generation assets
such as natural gas peaker plants. The majority of new
energy storage installations are using Li-ion batteries due
to their availability at large volumes and continued cost
reductions. In recent years, a lot of new manufacturing
capacity for Li-ion batteries has come online, primarily
geared for electric vehicles. This new manufacturing
capacity has led to significant cost reductions and helped
make Li-ion BESSs the technology of choice for grid ener-
gy storage systems.
It is important to note that electrochemical systems
with well-established manufacturing and quality-control
systems have very low failure rates, well below one in one
million, but large numbers of cells will likely be employed
in grid-scale energy storage systems. For example, a typical
18650 Li-ion cell has a capacity of 10 Wh, and to build a
1-MWh storage plant requires almost a hundred thousand
cells. Statistically, there is a significant chance of a single
cell failure occurring in such large systems. The number of
cells and system-level components associated with large-
scale energy storage systems leads to possibilities of a local
failure that is much greater than one in a million. Thus,
there is a need to provide system-level redundancies.
For large-scale energy storage with thousands of indi-
vidual cells arranged in packs and modules, there is an
additional concern with the propagation of a failure
20 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
event from one cell to multiple cells and possibly
through the entire system. This can occur through either
electrical or thermal transfer of energy. If a cell fails in a
manner that allows it to discharge its stored energy
through a short circuit, then electrical connectivity might
allow a much larger number of cells to also release energy.
This is particularly an issue for cells connected in paral-
lel to allow greater transfers of power. If a single cell
failure generates large quantities of heat, thermal propa-
gation of failure is possible as well. In these ways a single
point of failure can lead to a cascading series of failures
throughout the system.
With growing numbers of large battery fires involving
mostly Li-ion BESSs, there is a growing concern from
industry experts and regulators about potential safety
issues. Until recently, safety of energy storage systems
has been looked at almost exclusively through the nar-
row lens of cell-level failures. While a greater understand-
ing of cell-level failures has been critical to the success of
rechargeable batteries in consumer electronics, the com-
plexity associated with the scale of energy in grid applica-
tions necessitates consideration of a wider range of
system-level issues related to power electronics, to air
conditioning systems, and to fire suppression for large
energy storage systems and the surrounding physical
infrastructure. In addition, the kinetic behavior of cell-lev-
el failure must consider the probability of propagation
and thermal runaway that is not indicated in smaller bat-
teries of the same chemistry.
The need for a comprehensive system-level approach
to safety, training, and preparation for first responders to
handle BESS failures is considered a priority by National
Fire Protection Association (NFPA), International Associa-
tion of Fire Fighters (IAFF) and other agencies representing
first responders. Recent examples of thermal runaway
incidents with grid-scale storage facilities and electric
vehicles show that first responders are not adequately
prepared to handle these incidents.
Complicating matters further are new types of electro-
chemical energy storage technologies being connected to
the grid. These include Li-ion batteries with high energy
anode materials, Li-metal anodes, nonflammable electro-
lytes; flow batteries using a variety of electrolyte materials,
advanced aqueous batteries including advanced lead-acid,
alkaline batteries like rechargeable zinc batteries, and
molten salt batteries. Each system has its own hazards
that vary with chemistry and design that must be man-
aged. The industry likely does not have an adequate
understanding of the operational safety of these systems.
Grid-scale systems are inherently complex, requiring not
only a large battery but also sophisticated power electron-
ics, energy management systems, thermal management
systems, transformers, and grid interconnection. Integra-
tion of new battery types with these components is still in
progress, creating many potential points of failure. One
has to ask the question: what type of laboratory testing is
adequate to certify new systems for safety? Will small-
scale single cell testing, which may not simulate real-
world conditions, be sufficient or will full-scale testing,
which is prohibitively expensive, be necessary.
Engineering Safety
Safety engineering has a long history, and current indus-
trial practices in safety engineering were built, in part, out
of industrial or occupational safety requirements. Engi-
neering safety is to reduce risk of failures or mitigate
potential for accidents or incident with substantial effect
on life and property. In the energy storage domain, risk
management is a useful framework for considering engi-
neering/testing standards, along with siting/permitting
policy. Safety standards establish a list of falsifiable quali-
ties for an engineered system that, when checked, reduce
the risk of accidents. An example of this kind of standard
is the definition of size, separation, and maximum allow-
able quantity limits for different battery types, as is done
in the 2020 version of NFPA 855. Enforcing these limits,
with specific exceptions, constrains the feasible worst-
case-scenario by preventing a fire in one block of cells
from spreading.
Battery Thermal Runaway
All forms of stored energy can be a hazard if that energy is
released without control. Batteries are inherently remark-
able systems, as they remain inert until an external circuit
is connected, and only then release energy as electric cur-
rent. Battery failure occurs when an event, either internal
or external to the battery, causes this stored energy to be
released in an uncontrolled manner. If that release is fast
enough, heating can create a feedback loop of increasing
heat generation and subsequent combustion called ther-
mal runaway. During thermal runaway, the release of
stored energy is occurring so rapidly that intervention is
effectively impossible. Fire is typically described as a trian-
gle among fuel, oxidizer, and heat. In a charged Li-ion bat-
tery, the two electrodes effectively provide the fuel and
oxidizer within the cell, meaning the heat needed to initi-
ate a thermal runaway is the only item missing. While fire
(a)
Figure 1. The failure of a 1-kWh battery pack from overcharging. (a) Test setup. (b) Battery fire resulting from overcharge. (c) Post-test condition.
resulting from combustible materials might be extin-
guished through a variety of means, the only way to stop
the thermal runaway event is to remove the heat that is
sustaining it.
Broadly, single cell failures can be grouped into field
and abusive failure categories. Abusive failures are rela-
tively easy to identify as the failure is driven by the battery
being exposed to a condition it is not designed to with-
stand. This can include mechanical damage or exposure
to temperature, voltage, or current outside the battery’s
normal operating window. Field failures are more difficult
to predict and detect. They are roughly defined as an
unexpected failure while idle or during normal operation
of the battery, for example, an internal short circuit caused
by a contaminant in the cell.
The susceptibility of single cells or battery packs to
abusive failures as well as the potential consequences are
typically evaluated using standard industry tests (see Fig-
ures 1 and 2). Peak temperatures during failure, the condi-
tions where failure occurred, and video recordings of the
failure are all evaluated to determine the general severity
and susceptibility to abuse conditions. Taken as a whole,
this provides a view of the true safe operating envelope of
a particular battery as well as the worst-case scenario
during failure. This information can then be used to aid in
the development of a safe and reliable battery system.
The evaluation of field failures is more challenging as
statistical failure rates can vary by environmental and
operational conditions. Because these failures are typically
a rare and random process, it is difficult to evaluate the
risk of spontaneous failure of a grid connected energy
storage system. This is most easily evaluated in small,
commercial-off-the-shelf batteries from major manufac-
turers as the large numbers of cells produced enables esti-
mates of a true failure rate, which is from where the
previously mentioned less than one in one million failure
rate comes. The large cell formats used in grid-scale stor-
age systems are typically much more specialized and pro-
duced in relatively small quantities. This means
evaluating the vulnerabilities of grid storage cells to field
failure is difficult, and we are left with relying on battery
(b) (c)
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 21
 abuse testing to at least determine the worst-case out-
comes. This can ultimately be misleading, however, as sin-
gle-cell testing will not consider system-level mitigations,
nor identify any vulnerabilities in place due to poor
design. Some degree of system-level testing helps to
bridge some of the gaps in determining the full risk posed
by the system.
Full System-Level Fire Testing
In recent years, battery-related products for large-scale
energy storage have increased in capacity and energy
density at a remarkable rate. Serious incidents have
occurred with large grid-tied Li-ion BESSs, raising public
safety issues. In most cases, these kinds of incidents hap-
pen when thermal runaway occurs in a battery cell and
propagates to the surrounding batteries. The inherent
properties of Li-ion cells make it extremely difficult to
avoid thermal runaway in a single cell. Therefore, inter-
ruption of thermal runaway propagation to the surround-
ing cells and modules is extremely important for the
safety of large-scale BESSs.
It is very important to understand experimentally how
the propagation of thermal runaway occurs inside the
large-scale BESSs in actual environments. Based on the
experimental results in a controlled environment, final
BESS products can be better engineered to reduce the
threat of thermal runaway propagation.
There are only a few places in the world, such as
Underwriters Laboratories (UL) in the United States and
the National Institute of Technology and Evaluation
(NITE) in Japan, where abusive testing of large MWh-sized
Test Description
Mechanical crush of a large area of
Controlled Crush the battery; potentially performed in
multiple orientations
Penetration of a battery with a nail,
Penetration
spike, or similar object
Testing battery resilience to
overtemperature conditions; typically
Thermal Ramp performed by gradually increasing
ambient temperature until failure
occurs
Charging of a cell beyond its fully
Overcharge charged state; may include polarization
due to high voltage
Discharge of a cell beyond its normal
zero energy state; may include voltage
Overdischarge
reversal where battery polarity is
forcibly reversed
Application of an external,
External Short
low-resistance electrical load
Circuit
Figure 2. The typical battery abuse tests.
22 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
BESSs can be performed. NITE has investigated in detail
the propagation behavior of thermal runaway in a cabi-
net-level battery systems, utilizing the multipurpose test-
ing facility in NLAB (National LABoratory for Advance
Energy Storage Technologies) in Osaka, Japan. NLAB Large
Chamber is a huge indoor test facility for large-scale
BESSs, including containerized systems. In one experi-
ment, one module in a cabinet-scale test sample was
overcharged to cause thermal runaway. Then, the temper-
ature and voltage change in the cabinet was observed to
track the propagation of thermal runaway through the
cabinet. It was found that thermal runaway did not
always propagate through cells in direct contact with
each other. Observations after this experiment showed
that a large hole had been melted through the side plate.
Conflagration in a battery system located far away from
the initial point of heat generation is related to genera-
tion of flammable gases from the battery undergoing
thermal runaway. Thus, the suppression of fumes and arc
flash is critical to preventing propagation of thermal run-
away. This experiment highlights the new discoveries
enabled by large-scale safety testing. The results informed
the development of IEC62933-5-2 in 2020 as the first inter-
national standard to define safety requirements for sys-
tem-level BESSs.
Hazards from Battery Vent Gases
While significant attention is often paid to thermal run-
away events in batteries, the most severe consequences
have often been caused by the ignition of flammable
gases released during Li-ion battery failure. Ventilation of
spaces operating Li-ion batteries is often limited, as,
unlike many other battery technologies, there is generally
no gas release from Li-ion cells during normal operation.
However, during failure events, significant quantities of
gas can be rapidly released. A heuristic rule used within
Sandia’s test laboratories gives up to 2 L of gas released
per amp hour of battery capacity during a battery failure.
At these quantities, even a single cell failing can quickly
fill a room with vented gas. Typical constituents include
hydrogen, carbon monoxide, methane, and aerosolized
liquid electrolyte. These can reach flammable or even
explosive concentrations and be ignited by sources com-
mon within an energy storage system. Figure 3 shows
previously published data from Sandia of the decomposi-
tion products of common solvents containing 1 M LiPF6
(the typical salt used in battery electrolytes). While CO2 is
the most common component, this also shows signifi-
cant production of flammable gases. Figure 4 shows how
these gases and aerosolized electrolyte can be forcibly
ejected from a battery during failure.
Battery thermal runaway is typically accompanied by
rapid release of these gases; sparking or hot ejected mate-
rials from the battery provide a convenient ignition source.
The most catastrophic scenarios often involve an initial
venting of gases from the battery at lower temperature,
first filling an enclosed space. This is then followed up by a currently exhibit significantly lower current capability
thermal runaway event, generating yet more gas and pro- than traditional electrolytes, limiting their use for high
viding an ignition source for the built-up gases. power applications.
Recent scenarios of fuel-air ignition driving a cata- There are several leading system design strategies for
strophic failure are evident in battery systems fires that mitigating potential hazards from gas venting from Li-
occurred in the last few years. When first responders ion cells. The simplest strategy is to prevent thermal run-
entered the space, the accumulated gases ignited, result- away propagation as the gas vented from single cell
ing in an explosion that destroyed the container and failure is unlikely to be sufficient for hazardous combus-
injured several of the responding firefighters. Accident tion. As many battery systems need to be located in a
analysis reports suggest that the container was filled with temperature controlled, and hence enclosed space, the
flammable gases. While it is likely difficult to fully account next best strategy is to limit the maximum quantity of
for every scenario, this demonstrates the need for educa- batteries to below the threshold where their vented
tion on some specific hazards inherent to large battery gas would reach the low explosivity limit (LEL) in the
systems, particularly for system
installers and first responders.
Significantly, the primary source
for these gases is the breakdown of 100
the electrolyte necessary for the DEC
80
normal function of Li-ion batteries. EC
Research is ongoing to try and find 60 EMC
a replacement for this electrolyte, 40 DMC
but the potential solutions identi-
20
fied to date severely reduce the
power capability and lifetime of the 0 DMC
battery or are cost-prohibitive to O2 H2 EMC
C O EC
manufacture. A potential solution C H4 H6
C
C2 H8 10 2 DEC
C 3 H H 1 1 4 r
includes the use of all solid-state C4 H e e
C5 C6 th an
y le t h
electrolytes, which are particularly h e
Et ro
attractive as they have the potential l uo
F
to enable a lithium metal anode,
dramatically increasing the avail- Figure 3. The molar percentage of gas decomposition components detected from the thermal
breakdown of diethyl carbonate (DEC), ethyl carbonate (EC), ethyl methyl carbonate (EMC), and
able energy in batteries. However, dimethyl carbonate (DMC) all with 1 M LiPF salt. (Courtesy of F. Meier and M. Hargather, New
6
even the best solid-state electrolytes Mexico Tech.)

1
20 mm 10 mm 10 mm
0.9

0.8

0.7
Time of Arrival (ms)

0.6

0.5

0.4

0.3

0.2

0.1

0
(a) (b) (c)

Figure 4. Image shows time of arrival of vented material from an 18650 cell after failure: (a) gas phase plume, b) low viscosity (2.57 cP) elec-
trolyte ejection, and c) high viscosity (329 cP) electrolyte ejection.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 23

 enclosed space. Lastly, if this quantity limit cannot be
feasibly enforced, one proposed design strategy is to acti-
vate emergency ventilation when explosive gases are
detected. This needs to be done consistent with NFPA 69,
at a flow rate sufficient to prevent gases from reaching
their LEL. If all else fails, enclosures that could contain a
potentially explosive gas can be designed with deflagra-
tion panels, consistent with NFPA 68, to direct an explo-
sion in a nonhazardous direction.
Responding to a Li-ion BESS Fire Event
Imagine a firefighter, arriving at a BESS site with smoke
rising from cracks and corners. In this situation they
would start to assess the potential hazards: is anyone in
danger, could the fire spread, what kind of batteries are
involved? Those involved in the design, integration, and
approval of battery installations should be working to
make this risk assessment as easy and simple as possible.
Firefighters should be able to access the information
required to assess risk and should have the training need-
ed to interpret this information. Safety critical information
might include: what system chemistry is present, what
percentage of the cells in the system have vented, is the
ventilation system working as expected, what voltages are
present in the system, are the batteries getting hotter or
colder and by how much, what actions have been taken by
the automated systems (e.g., fire suppression), or if there
are any hazardous gases present. Much of this informa-
tion can be made available through a fire alarm control
panel that would be placed in a safe location away from
the battery. Table 1 shows a summary of available and
upcoming training resources for first responders in the
United States. Similar training is becoming available in
Europe, Japan, and other locales.
An existing course developed by the National Fire Pro-
tection Agency offers guidance on risk assessment in
energy storage fires such as identifying ac and dc electri-
cal hazards, common failure modes, shutdown, ventila-
tion, and other emergency response considerations.
Courses under development focus on both firefighters and
first responders who are building officials. All of the pro-
grams emphasize the importance of preincident planning
and working with local first responders to make sure that
they are aware of the installation and know what to do in
case of a fire.
One of the hardest aspects of responding to a battery
fire, primarily in Li-ion batteries, is knowing when it is safe
to release the equipment back to its owner and leave the
scene. The energy stored within a damaged battery has
the potential to reignite, hours or possibly days, after it has
been previously extinguished (“stranded energy”). The pre-
incident planning should take this possibility into account
and plan for a fire watch, as needed. A modular system
may be restored to operation shortly after a fire, if a fire is
confined to a single module or rack, though often with
reduced performance. However, smoke and fire suppres-
sion could necessitate significant cleanup or refurbish-
ment before a system can be recommissioned. It is
important to understand that many systems are being
designed to handle a battery cell fire gracefully and that
after some cleaning, part replacement, inspections, and
testing, these systems may continue to operate.

TABLE 1. The existing and upcoming national-level training for first responders in the United States.

Lead Organization Target Audience Description

NFPAa Firefighter Self-paced online energy storage and solar systems safety training, fire service edition;
3-h module uses engaging videos, animations, simulations, and review exercises to
inform firefighters about basic electrical theory, types of photovoltaic installations,
battery chemistries, and response strategies
NFPAb Firefighter Develop a suite of training tools, including a multiplayer serious gaming platform
with several energy-related incidents at each scene
IAFFb Firefighter Develop training programs by testing installed residential energy storage systems to
understand what happens when batteries fail and pose risks to emergency responders
IRECb Building officials Introduce educational materials and resources on clean energy codes, standards, per-
mitting, and inspection for those interacting with solar energy and storage systems
Southface Energy Building officials Create educational programming and resources that can be tailored to different
Instituteb markets and jurisdictions
New Buildings Building officials Develop guides and education modules to streamline the design, permitting, inspec-
Instituteb tions, and maintenance of solar, storage, and electric-vehicle charging stations for
single- and multifamily homes and offices
aExisting training.
bTraining under development (as of December 2020).

IREC: International Renewable Energy Council.

24 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

Aqueous Battery Systems— It is important to
Safety Aspects
Currently, Li-ion batteries are the understand that
dominant electrochemical energy
storage devices on the grid, as such, many systems are
they have received the bulk of atten-
being designed to
tion for hazard mitigation through
codes, standards and safety training. handle a battery cell
However, the hazards from other bat-
tery systems, particularly aqueous fire gracefully and
based systems, need to be considered that after some
as they are increasingly utilized for
grid applications. Aqueous batteries cleaning, part
are defined as batteries that use
water as their supporting electrolyte. replacement,
This broad category includes chemis- inspections, and
tries ranging from classic lead–acid to
newly designed rechargeable zinc- testing, these
manganese oxide with formats span-
ning typical cylindrical cells to redox
systems may
flow batteries (RFBs). Aqueous batter- continue to operate.
ies are generally safer than Li-ion sys-
tems given their more stable aqueous
electrolyte versus a flammable organ-
ic electrolyte and active species that are generally relative-
ly stable transition metals. There are still inherent risks for
aqueous batteries that must be managed. Safety concerns
for aqueous batteries will vary based on their chemistry
and design. The three primary risks are gas evolution,
electrolyte leakage, and thermal runaway.
Gas evolution can occur through different routes that
depend on the system and how it is operated. In aqueous
systems, hydrogen evolution can occur at all anodes
through electrochemical mechanisms. The rates and occa-
sion of hydrogen generation changes with the chemistry
and operational conditions. Hydrogen presents an explo-
sion and flame hazard if it builds up beyond 4% of the gas
mix. However, there are well-documented ways to man-
age the risk since hydrogen evolution in lead-acid batter-
ies has been studied for decades. Other gases can evolve
depending on the system in question. Most notably,
mixed-acid electrolyte vanadium RFB systems (electrolyte
contains sulfuric and hydrochloric acid) can produce Cl2
gas and zinc-bromine hybrid RFBs can produce Br2 gas
from their respective catholytes. Both gases are highly
toxic and corrosive, presenting new safety considerations.
For example, Cl2 gas has a 60-min exposure limit of 3 ppm
and an exothermic reaction with H2 gas that can be initi-
ated by a minimal ignition source.
The risk of electrolyte leakage depends on the sys-
tem. So-called starved systems have minimal amounts
of electrolyte, reducing the leak risk, but flooded batter-
ies and RFBs contain significant amounts of electrolyte
that can leak. The electrolyte for these systems are often
highly concentrated acids or bases and very caustic.
Consequently, primary containment needs to be made
from corrosion resistant materials
that can tolerate the electrolyte for
the lifetime of the system. Addition-
ally, secondary containment needs to
be put in place to hold any spill or
leaks that may occur. Loss of con-
tainment may occur through buildup
of pressure from gas evolution, or an
explosive reaction from accumulat­­
ed gases. Secondary containment
should be able to withstand such
forces, so electrolyte is not leaked
into the surrounding area. An exam-
ple of a loss of containment scenario
is shown in Figure 5.
Thermal runaway is a common
risk for lead-acid and other aqueous
batteries like nickel metal hydride and
nickel cadmium systems. RFBs and
zinc-based systems have yet to show
thermal runaway risks but cannot be
definitively ruled out as a possibility.
As with Li-ion systems, thermal run-
away in aqueous batteries involves
excessive heat generation, gas evolution, and even explo-
sions. Due to their more stable electrolyte and lower ener-
gy density, these events are typically less energetic than in
Li-ion systems, but they still can pose significant risks. A
recent NFPA report highlighted 14 significant events from
stationary lead-acid systems since 1993 that resulted in
damage to the surroundings, significant injury to person-
nel, and one notable case where a battery fire ignited a
methane pocket in a mine that led to 13 deaths
For all of the risks described previously, the scale and
system design will determine the potential impact of an
event. In a large bank of lead-acid or other nonflow aque-
ous batteries, thermal runaway may be able to ignite other
cells, like how Li-ion thermal runaway can propagate.
Adding more cells to the system increases the potential
size of the resulting fire or electrolyte leak. Grid-scale RFBs
consist of large tanks of electrolyte (tens of thousands of
gallons per MWh) connected to an electrode stack. Conse-
quently, significant amounts of gas can be generated
quickly in these systems, which can cause an explosive
reaction leading to loss of electrolyte containment. The
resulting electrolyte spill could involve thousands of gal-
lons of hazardous material.
Nonbattery ESS—Safety Aspects
Basic safety evaluations are still critical even from nonbat-
tery energy storage technologies that have been around
for decades. Here, we consider pumped-storage hydro-
power (PSH) and flywheels as examples. PSH plants store
and generate energy by moving water between two reser-
voirs at two different elevations. During times of low elec-
tricity demand, excess energy is used to pump water to an
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 25
 upper reservoir and during periods of high demand, the
storage water is released through turbines to generate
electricity. In 2020, PSH accounted for over 90% of active
storage installations worldwide, with a capacity of
1.6 TWh. A review of thousands of incidents from the
Electric Power Development Company of Japan (J-Power), a
major designer and operator of PSH, found that almost all
issues with such facilities stemmed from machinery fail-
ures of the pump turbine and generator motor. J-Power
determined 37% of incidents that caused mechanical
breakdown and/or outages lasting more than 10 days
were related to design or manufacturing flaws, and 58%
were due to deterioration or poor maintenance. The
review offered excellent recommendations for mainte-
nance best practices at each stage of the decades-long
operational lifetime of a PSH.
In the last few years, several maintenance issues have
cascaded into catastrophic failure of a PSH facility or loss
of life (Figure 5). In 2005, the upper reservoir of the Taum
Sauk Pumped Storage Plant in Missouri was overtopped
due to a faulty gauging system. Due to this and the lack of
a spillway, a large section of the upper reservoir failed and
over a billion gallons of water was drained in less than a
half hour. There were five injuries and power generation
did not resume for five years. In 2007, maintenance work-
ers at the Cabin Creek Generating Station in Colorado
were reapplying an epoxy lining to the inside of a sluice
pipe and using a low flash point solvent to clean equip-
ment. A flash fire broke out and five workers died, trapped
in the pipe because the fire was blocking the only egress
route. In 2020, a fire started in a control panel during
maintenance at the Srisailam Left Bank PSH in Telangana,
India. Several personnel were able to escape the four-story
powerhouse, but nine others were trapped and died before
rescue personnel arrived. Although PSH has served as a
critical grid energy storage technology for over 50 years,
each project still merits the same degree of safety scrutiny
as emerging technologies.
Flywheels operate by accelerating a rotor to a very
high speed and maintaining the energy in the system as
rotational energy. A typical system consists of a flywheel
supported by a bearing connected to a motor generator,
often placed in a vacuum chamber to reduce friction and
energy loss. Flywheels have been used for over a century
in applications including rail electrification, uninterrupt-
ible power supplies, and nuclear fusion research facilities.
In grid energy storage, they have been used for frequency
regulation and paired with wind turbines. Two key haz-
ards for a flywheel are rotor failure and a loose rotor.
Rotor failure occurs when the stress on a rotor exceeds its
strength. Failure can be mitigated by using an adequate
design margin (2X as required by UL9540) determined by

Exchange Membrane

Side Reaction in H2 Cl2

Cathode
Anode

Individual Cells Side Reaction H2 Cl2 Side Reaction

During Charging 2H++2e− => H2 H2 Cl2 2Cl−-2e− => Cl2
H2 Cl2 • 7% of Head Space Can
be Cl2 During Cycling
Anolyte Tank Electrolyte Flow Catholyte Tank
• In a 1-MW/4-MWh
H2 H Head Space Head Space Cl2 Cl2 Cl2 Mixed-Acid System this
2 is ~2,100 L of Cl2 gas
H2 H2 Cl2
H2 Cl2 • 17.5 MJ of Energy
Released If All the Cl2
Gas Is Consumed

Electrolyte Electrolyte

Cl2 H H Cl2 Cl2

2 2 Cl H2 H2 Cl
2 2
Cl2 H2 Cl2 H2 Cl2
H2 Cl2 H2 Cl2
Cl2 Cl2 H2

Electrolyte Electrolyte Electrolyte

Gases Mix in Head Space Catalyst (Spark, UV Light, etc) Exothermic Reaction
Initiates Reaction: Occurs Leading to Loss
Cl2 + H2 => 2HCl + Heat of Containment

Figure 5. The loss of electrolyte containment scenario from gas evolution and build up in a mixed-acid vanadium RFD system.

26 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

appropriate materials certification and testing. The con-
sequence of rotor failure is the production of energetic
and destructive projectiles. A loose rotor can be caused by
the failure of a bearing, shaft, or hub and results in dam-
age to the surrounding areas. These failure modes have
resulted in two public safety incidents (Table 2).
Ensuring Safety—Codes and Standards
Codes and standards are used to determine minimum
requirements of hazard mitigation for a system. There is a
hierarchical web of standards that impose requirements
on the design of energy storage systems at each level of
integration (Figure 6). Component standards such as UL
1741, on Li-ion batteries, or CSA C22.2 No. 340 on battery
management systems have checklists of testable and veri-
fiable metrics that must be met to be listed. Integrated
system standards like UL 9540 then place requirements
on how the components are connected together in a sys-
tem. Installation standards such as NFPA 855, Standard
for the Installation of Stationary Energy Storage Systems,
then specify where and how energy storage systems can
be installed safely. IEEE has standards that cover installa-
tion for specific battery chemistries (for example, IEEE 484
for vented lead-acid) and how to manage electrolyte spills
(IEEE 1578), which combined cover most hazards associat-
ed with aqueous batteries. Lastly, standards for the built

TABLE 2. Examples of safety incidents in nonbattery energy storage systems.

Technology Location Description

Flywheel October 2011; Stephentown, Two out of 200 3,000 pound flywheels failed due to flawed early production
Massachusetts, USA runs of their carbon fiber material. Flywheels ground against and heated the
chamber. Cooling water caused steam explosion.
Flywheel December 2015; Poway, Loose 11,000 pound flywheel crashes into vault. Warehouse deemed unsafe.
California, USA Four workers injured by debris.
PSH December 2005; Taum Sauk Upper reservoir overtopped and damaged due to faulty gauging system and
Pumped Storage Plant, lack of spillway. Over a billion gallons of water drained in less than half an
Missouri, USA hour, injuring five people and halting power generation for five years.
PSH October 2007; Cabin Creek Highly volatile solvent (methyl ethyle ketone) used during maintenance of
Generating Station, Georgetown, epoxy on pipe interiors. Flash fire trapped and led to deaths of five workers.
Colorado, USA
PSH August 2020; Srisailam Fire started in control panel in powerhouse during maintenance. Nine
Left Bank, Telangana, India workers died

All Buildings in a Specific Area

Built Environment
4 International Codes–IFC, IRC, IBC
IEEE–C2, SCC 18, SCC 21
NFPA 5000, NFPA 1, ISA

Installation/Application
NFPA 855, NFPA 70, IEEE C2, IEEE
3
1635/ASHRAE 21, IEEE P1578,
FM Global 5-33, UL 9540A, NECA 416
Installed and
Energy Storage Systems Integrated
Systems
2 UL 9540, MESA
ASME TES-1, NECA
NFPA 791

System Components
1 UL 1973, UL 1974, UL 810A, UL 1741,
CSA 22.2 No. 340-201, IEEE 1547, IEEE 1679 Battery Cells,
Integrated System Modules, Inverter,
(Not Installed) BMS, ESMS, etc.

Figure 6. The rough categories for energy storage codes and standards.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 27

 environment, such as the International Fire Code (IFC),
provide general requirements for all buildings. While
these are imperfect categories, as individual standards
often span several aspects of the system, they illustrate
that requirements for a specific installation can be found
in a broad range of applicable standards.
The last few years has seen rapid development in safe-
ty standards for energy storage technologies. The IFC cov-
ers electrical energy storage systems, and its requirements
were updated in the 2021 version. NFPA 855 was published
in 2019 and revised in 2020 to accommodate more battery
types. IEC 62933-5-2 ED2 Safety requirements for grid-inte-
grated EES systems was published in 2020. CSA C22.2 No.
340 was just completed and published in 2021. However,
despite how quickly the relevant codes and standards are
being updated, they will continue to lag the cutting edge
of technological development.
Technology standards are developed through consen-
sus processes with diverse stakeholders and tend to lag
technology deployment. For example, despite the com-
bustion hazard from battery vent gases discussed earlier
in this article being well understood from laboratory
experiments, events related to vented gases have injured
first responders. However, it was partly this event that
spurred standards development working groups to incor-
porate analyses of the hazard into their revision process-
es. The forthcoming 2021 version of IEEE 1635/ASHRAE
21, Guideline for the Ventilation and Thermal Management of
Batteries for Stationary Applications, will include informa-
tion on the combustion hazard from battery vent gases
and guidance for appropriate heating, ventilation, and air
conditioning design for Li-ion battery systems. The stan-
dard development and revision processes are very good
at learning from historical accidents, but they can fall
short trying to prevent accidents before they happen.
Hence both standards and safety engineering have roles
to play in keeping energy storage systems safe.
Summary
Grid energy storage technology is at the cusp of major
growth with rapidly accelerating pace of deployments.
Energy storage, especially BESSs, is a new technology for
grid operators. There is still a lot of groundwork that needs
to be done to enable rapid adoption of energy storage in the
grid. We need to adequately address current gaps in safety
research and develop robust safety standards and protocols
for safe operation. We also need to make sure stakeholders
are adequately informed of the risk they can accept and
are fully prepared to deal with any potential failures. It is
also important to provide first responders with adequate
training and resources needed to handle incidents.
Acknowledgments
This work was supported by the U.S. Department of
Energy, Office of Electricity, Energy Storage program.
Sandia National Laboratories is a multimission
28 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
laboratory managed and operated by National Technolo-
gy & Engineering Solutions of Sandia, LLC, a wholly
owned subsidiary of Honeywell International Inc., for
the U.S. Department of Energy’s National Nuclear Securi-
ty Administration under contract DE-NA0003525. This
article describes objective technical results and analysis.
Any subjective views or opinions that might be
expressed in the article do not necessarily represent the
views of the U.S. Department of Energy or the United
States Government.
For Further Reading
J. Lamb and J. A. Jeevarajan, “New developments in battery
safety for large-scale systems,” MRS Bull., vol. 46, no. 5, pp.
395–401, 2021. doi: 10.1557/s43577-021-00098-0.
“BESS failure event database,” Electric Power Research
Institute. https://storagewiki.epri.com/index.php/BESS_Failure
_Event_Database (accessed Aug. 17, 2021)
M. Yasuda and S. Watanabe, “How to avoid severe inci-
dents at hydropower plants,” Int. J. Fluid Mach. Syst., vol. 10, no.
3, pp. 296–306, July/Sept. 2017. doi: 10.5293/IJFMS.2017.10.3.296.
D. Rosewater et al., “Grid-scale energy storage hazard
analysis & design objectives for system safety,” Sandia
National Laboratories Report, Albuquerque, NM, SANDIA Rep.
SAND2020-9360, Aug. 2020.
“McMicken battery energy storage system technical analy-
sis and recommendations,” Arizona Public Service, Phoenix,
AZ, Final Rep. DNV GL Document # 102090302-HOU-R-01, July
18, 2020. Accessed: Aug. 17, 2021. [Online]. Available: https://
www.aps.com/en/About/Our-Company/Newsroom/Articles/
Equipment-failure-at-McMicken-Battery-Facility
T. Parker, L. Obeng, and Q. Wang, “Fire hazard assessment
of lead-acid batteries,” Fire Protection Research Foundation of
the NFPA, Quincy, MA, July 2020. [Online]. Available: https://
www.nfpa.org/-/media/Files/News-and-Research/Fire-statistics
-and-reports/Hazardous-materials/RFLeadAcidBattery.pdf
R. M. Wittman, M. L. Perry, T. N. Lambert, B. R. Chalamala,
and Y. Preger, “Perspective—On the need for reliability and
safety studies of grid-scale aqueous batteries,” J. Electrochem.
Soc., vol. 167, no. 9, p. 090545, 2020. doi: 10.1149/1945-7111/
ab9406.
Biographies
Babu R. Chalamala (bchalam@sandia.gov) is with Sand-
ia National Laboratories, Albuquerque, New Mexico,
87185, USA.
David Rosewater (dmrose@sandia.gov) is with Sandia
National Laboratories, Albuquerque, New Mexico, 87185, USA.
Yuliya Preger (ypreger@sandia.gov) is with Sandia
National Laboratories, Albuquerque, New Mexico,
87185, USA.
Reed Wittman (rwittm@sandia.gov) is with Sandia
National Laboratories, Albuquerque, New Mexico,
87185, USA.
Joshua Lamb (jlamb@sandia.gov) is with Sandia Nation-
al Laboratories, Albuquerque, New Mexico, 87185, USA.
Akira Kashiwakura (kashiwakura-akira@nite.go.jp) is
with the National Institute of Technology and Evaluation,
Osaka, 559–0034, Japan.

 By Jim McDowall, Matthew Paiss, and Michael Ropp

The Codes and

Standards Facilitating
the Design and Adoption
of Energy Storage for
Power System Applications
Keeping pace with evolving safety codes and standards.

NERGY STORAGE, PRIMARILY IN THE FORM How we arrived here was not so easy. The early sys-

E of lithium-ion (Li-ion) battery systems, is

growing by leaps and bounds. Analyst
Wood Mackenzie forecasts nearly 12 GWh
of deployments in 2021 in the United
States alone. Installations of more than 100 MW and hun-
dreds of megawatthours are becoming commonplace.
tems were designed and installed largely in the absence
of safety standards and codes, with manufacturers and
system integrators taking their individual approaches to
safety. Missteps have been made, as indicated by the Elec-
trical Power Research Institute’s Battery Energy Storage
System Failure Event Database (storagewiki.epri.com),

©SHUTTERSTOCK.COM/PETRMALINAK
Digital Object Identifier 10.1109/MELE.2021.3115556
Date of current version: 1 December 2021

2325-5897/21©2021IEEE IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 29

 1,400
Residential
1,200 Nonresidential
Front of the Meter
1,000
Deployments (MWh)

800

600

400

200

0
Q1 2013
Q2 2013
Q3 2013
Q4 2013
Q1 2014
Q2 2014
Q3 2014
Q4 2014
Q1 2015
Q2 2015
Q3 2015
Q4 2015
Q1 2016
Q2 2016
Q3 2016
Q4 2016
Q1 2017
Q2 2017
Q3 2017
Q4 2017
Q1 2018
Q2 2018
Q3 2018
Q4 2018
Q1 2019
Q2 2019
Q3 2019
Q4 2019
Q1 2020
Q2 2020
Q3 2020
Q4 2020
Q1 2021
Q2 2021
Figure 1. U.S. deployments of battery energy storage by quarter. (Source: Wood Mackenzie Power Renewables/U.S. Energy Storage Monitor;
used with permission.) Q1: first quarter; Q2: second quarter; Q3: third quarter; Q4: fourth quarter.

which includes 41 events in systems totaling at least This article describes the development of ESS-related
228 MWh, since 2017. codes and safety standards, with a focus on North Ameri-
Codes and standards have been playing catch-up, in ca, and their impact on the industry. It will also detail stan-
some cases with multiple iterations, since 2017. The dards that go beyond the realm of safety, making it easier to
result is the imposition of a certain discipline—albeit a evaluate new storage technologies and interconnect to the
rather costly one—on the integrators of energy storage electric power system.
systems (ESSs) in their designs and deployments.
Although the development of these codes and standards ESS Deployments
is by no means complete, the power industry can pro- Throughout the 1990s and most of the 2000s, nearly all
ceed with new ESS deployments with a greater sense of large-scale, grid-connected energy storage was in the form
confidence in their safety. of pumped-hydro installations, and the systems using bat-
tery storage were mostly limited to
funded demonstrations. That start-
ed to change in 2009 with the first
100
shipments of containerized Li-ion
90 batteries, and by the late-2010s, the
80 deployment of battery-based ESSs
70 had entered a period of exponential
60 growth, as depicted in Figure 1.
(%)

50 Although the figures for the first

40 quarter of 2021 showed a significant
30 drop from the previous quarter,
20 Wood Mackenzie described this peri-
od as the “quiet before the storm,”
10
forecasting a total of nearly 12 GWh
0
to be deployed for the full year.
Q1 2013
Q2 2013
Q3 2013
Q4 2013
Q1 2014
Q2 2014
Q3 2014
Q4 2014
Q1 2015
Q2 2015
Q3 2015
Q4 2015
Q1 2016
Q2 2016
Q3 2016
Q4 2016
Q1 2017
Q2 2017
Q3 2017
Q4 2017
Q1 2018
Q2 2018
Q3 2018
Q4 2018
Q1 2019
Q2 2019
Q3 2019
Q4 2019
Q1 2020
Q2 2020

The share of those deployments,

by technology, is presented in Fig-
ure 2. Recent installations are made
Li-ion Other
almost exclusively with Li-ion
chemistries, and indeed, Li-ion has
Figure 2. The Li-ion share of U.S. energy storage deployments. Q1: first quarter; Q2: second
quarter; Q3: third quarter; Q4: fourth quarter. (Source: Wood Mackenzie Power Renewables/U.S. constituted the majority of deploy-
Energy Storage Monitor; used with permission.) ments since the early 2010s.

30 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

Li-ion Safety Issues While the industry
All those early installations of Li-ion
ESSs were made in the absence of understanding of
safety standards relevant to the scale
of these systems. At that time, the holistic safety
standards applicable to portable bat- has been advancing,
teries typically covered safety at the
level of individual cells and small codes and standards
packs, while the issues facing larger
systems were more concerned with have been playing
the propagation of thermal runaway catch-up.
(TR) between modules and racks,
effectiveness of fire suppression, and
management of explosive gas build-
up. It is not the purpose of this article to describe the
details of these safety issues but to show how current
codes and standards are i­nfluencing Li-ion system design
and providing both the  operators and authorities having
jurisdiction (AHJ) with a greater sense of confidence in the
deployment of these systems.
Safety Standards
Development Timeline
In those earlier ESS deployments, it was left to the bat-
tery manufacturers and system integrators to ensure an
adequate level of safety. Inevitably, there have been mis-
steps along the way, with system designers not always
considering all the ways in which cell failure can influ-
ence system-level safety. While the industry understand-
ing of holistic safety has been advancing, codes and
standards have been playing catch-up. The standards
from Underwriters Laboratories started to move beyond
the portable battery world in 2013, but it was not until
the last few years that they comprehensively addressed
complete battery-based ESSs, with the second edition of
UL 1973, Standard for Batteries for Use in Stationary, Vehicle
Auxiliary Power and Light Electric Rail (LER) Applications
(2018); the second edition of UL 9540, Energy Storage Sys-
tems and Equipment (2020); and the fourth edition of UL
9540 A, Standard Test Method for Evaluating Thermal Run-
away Fire Propagation in Battery Energy Storage Systems
(2019). Fire codes have similarly been evolving, introduc-
ing stringent spacing and energy limitations for indoor
systems beginning in 2018. The code requirements for
ESSs have been further developed in NFPA 855, Standard
for the Installation of Stationary Energy Storage Systems
(2020), which is already in the process of being updated,
supported by more data from the field.
One of the challenges for manufacturers is product
development in the midst of an evolving landscape of
installation and product safety standards. The key
model fire codes in the United States, the International
Fire Code and NFPA1 Fire Code, are amended on a three-
year cycle. The technology is evolving more rapidly than
the codes cannot keep up with evolution of the
technology, which may unintendedly
create barriers to solutions not
addressed or impose restrictions
with limited justification. The chang-
es in product safety standards can
require recertification, which may
impose significant time delays and
cost. It is imperative to have broad
stakeholder input into the develop-
ment of these consensus codes and
standards to ensure that the best
possible language is achieved with
the most current data available.
Mind the (Safety) Gap
An example of just one of the challenges fire codes
attempt to address is in the area of fire suppression and
explosion control. The current requirements call for fire
sprinklers for many systems installed indoors (which
includes walk-in enclosures) and outdoors near expo-
sures. Historically, sensitive electronics have been well
protected by clean-agent suppression systems, such as
Novec 1230, FM200, or carbon dioxide. As these agents are
designed to either displace oxygen or rapidly cool the
environment, they are quite effective for visible flames,
but for certain Li-ion chemistries are ineffective in sup-
pressing deep-seated fires in modules due to exothermic
and oxide-releasing reactions during TR.
When a clean-agent system is designed, one require-
ment is to maintain a certain concentration and duration
in the space. This requires that the area be sealed from
exterior air, and as a result may allow continued TR and
the release of highly flammable gas constituents, such as
hydrogen, carbon monoxide, and hydrocarbons. So, in
essence, the attempt to suppress a fire will often create an
explosion hazard. Continued research is needed in effec-
tive suppression systems designed to address fires at the
module level. Manufacturers must consider the explosion
hazard and respond with mitigation designs.
Similarly, it is possible for water-based fire-suppression
systems (FSSs) to cause more problems than they solve.
Water has much greater cooling efficiency than clean
agents, but it must be supplied in sufficient volume within
and around battery modules to arrest the propagation of
TR, thus limiting cell venting and the release of flammable
gases. Ceiling-level sprinklers may not be effective at deliv-
ering sufficient water if modules are tightly packed, and
this may favor designs with more directed water flow
across the tops of modules. Another potential issue is with
water-mist systems, which, depending on the ingress pro-
tection rating of the battery modules, may deliver enough
water flow to create short circuits between cells but not
enough to remove the heat generated by those shorts. The
need for, and effectiveness, of FSSs with Li-ion batteries
can be demonstrated most effectively by large-scale testing
according to UL 9540 A, as discussed later in this article.
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 31
 System Certification
Since 2017, the National Electrical
Code has required batteries other
than lead acid to be listed, that is, test-
ed by a nationally recognized testing
laboratory to comply with an applica-
ble safety standard. For stationary bat-
teries, that standard is UL 1973, first
published in 2013 and extensively
revised in 2018. Additionally, batteries
such as Li-ion, which incorporate a
battery management system with
safety functions, must demonstrate
compliance with UL 991, UL Standard
for Safety Tests for Safety-Related Controls
Employing Solid-State Devices [Third Edi-
tion, (2010)] and UL 1998, UL Standard
for Safety Software in Programmable
Components [Third Edition (2013)].
Certification is not limited to the battery, however, as
the complete ESS must be certified to UL 9540, first pub-
lished in 2016 and revised in 2020. Certification to UL
9540 requires that a UL 1973-listed battery be paired with
a UL 1741-listed power-conversion system, with addi-
tional testing to verify the adequacy of system integra-
tion. The hierarchy of codes and standards applicable to
ESSs is depicted in Figure 3.
Large-Scale Fire Testing
Arguably the most impactful, safety-related document for
batteries is UL 9540 A, which is not a qualification stan-
dard but rather a standardized test method with no pass/
fail criteria. This testing assesses the extent of TR
Standards and Model Codes Hierarchy
Built Environment
• iCodes - IFC, IRC, IBC
• IEEE - C2, SCC 21
• NFPA 1, 5000
Energy Storage Systems
• NFPA 855 • FM GLOBAL 5-33
• UL 9540 • DNVGL GRIDSTOR
• ASME TES-1 • NECA 416 and 417
Installation/Application
• NFPA 70 • IEEE C2
• UL 9540 A • IEEE 1635/
ASHRAE 21
• IEEE P1578
System Components
• UL 1973 • UL 1741
• UL 1974 • CSA 22.2 NO. 340-201
• UL 810A • IEEE P2686
Figure 3. The hierarchy of ESS codes and standards. IFC: Interna-
tional Fire Code; IRC: International Residential Code; IBC: Internation-
al Building Code. (Source: Pacific Northwest National Laboratory;
used with permission.)
32 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
Arguably the most propagation within the battery and
the effectiveness of fire suppression,
impactful, safety- if needed. UL 9540 A testing is refer-
enced by fire codes and NFPA 855 and
related document for is required whenever it is proposed to
batteries is UL 9540 exceed mandatory energy limits or to
reduce mandatory spacing between
A, which is not a battery segments, or between batter-
ies and exposures. Results of the test-
qualification ing, including video recordings, are
standard but rather submitted to the AHJ to inform their
decision making on code variances.
a standardized test The evolution of UL 9540 A is an
example of the response of the stan-
method with no dards-writing community to unfold-
pass/fail criteria. ing ESS safety events around the
world and the lessons learned from
those events. Originally published in
2017, the standard was revised twice
in 2018, and most recently, the fourth edition was pub-
lished in November 2019.
The test method in UL 9540 A takes a stepwise
approach, with testing starting at the level of individual
cells and continuing to modules, units (racks), and finally,
complete installations. The document is applicable to all
battery technologies that are subject to installed-energy
and spacing restrictions in fire codes or in NFPA 855, and
there are “off ramps,” where systems giving satisfactory
results at one of the lower levels do not have to complete
the higher, more expensive levels. However, if TR or flam-
mable gas is generated at the cell-level test, the module-
and unit-level tests are required.
At each level, TR is initiated using an external heater. If
there is venting in the single-cell test, which is normal for
most Li-ion technologies, the overall volume and constitu-
ents of the vented gases are measured. For the module-
level test, a single cell is heated to TR, and the adjacent
cells are observed to understand the effects of propaga-
tion of the event. In the absence of sustained propagation,
additional cells may be heated to create a more extreme
event. There has been some discussion of this methodolo-
gy as the most likely initiating event in real life is an inter-
nal short in a single cell, and simultaneous internal shorts
in multiple adjacent cells would be statistically highly
improbable. The counterargument is that such multicell
• IEEE 1547 heating could potentially be caused by an arcing event.
• IEEE 1679 The testing moves on to the unit or rack level, with TR
initiation in one module and monitoring for possible mod-
ule-to-module propagation and measurement of vented
gases and radiated heat. If there is no flaming outside of
the unit or deflagration risk, testing can be terminated at
this stage.
Testing up to this point is done without fire suppres-
sion, so a lack of propagation in these tests indicates a
good level of intrinsic safety. If flaming occurs outside the
unit or gas generation represents a deflagration hazard,
the final installation-level test is performed with fire sup-
pression and other safety measures in the design to
assess the effectiveness of preventing unit-to-unit propa-
gation and preventing or reducing deflagration risks.
Conducting tests according to UL 9540 A can be time-
consuming and expensive, particularly if installation-level
testing is required. However, the wealth of information
provided, including video recordings, is invaluable in
allowing the AHJ to make informed decisions on proposed
installations and for fire protection engineers to design
fire-suppression and/or deflagration systems.
Influence of Safety Standards on Battery Design
As ESS deployments have shifted to multihour applica-
tions such as the replacement of peaking generation, a
system’s footprint has become progressively more impor-
tant, causing containerized Li-ion battery systems to be
designed with increasing energy density. Figure 4 shows
the rise in energy density over time for containerized Li-
ion battery systems offered by one company: Saft. All of its
systems use Li-ion chemistry based on nickel oxides,
including Li nickel cobalt aluminum oxide and lithium
nickel cobalt manganese oxide (NMC).
The earlier container designs included relatively small
modules that incorporated thermal barriers between large
cylindrical cells, and those barriers were effective at pre-
venting cell-to-cell propagation of TR. A standard FSS
using a clean agent, Novec 1230, is installed to handle any
electrical fires that may occur, and its cooling effect pro-
vides an additional safety margin against propagation. For
the 2020 container design, the installed energy is maxi-
mized using close-packed pouch cells in large modules.
This arrangement makes it difficult to avoid propagation
of TR. A clean-agent FSS extinguishes any visible flame
but is unable to eliminate ongoing propagation deep with-
in the modules. In this case, the primary FSS is supple-
mented by a water FSS.
The easiest way to implement a water FSS is to have
a dry-pipe system with a connection for the fire
(a)
Figure 4. The evolving energy density in 20-ft containerized systems. (Source: Saft America Inc.; used with permission.) (a) 2012: 0.6 MWh, (b)
2017: 1.2 MWh, and (c) 2020: 2.5 MWh.
department and to provide firefighters with status infor-
mation and a decision tree for when to use the system.
Ideally, the connection point would be located a safe dis-
tance away, although that may create issues for service
access to the containers. Moreover, a manually actuated
system may be problematic from an effectiveness stand-
point as it requires human intervention within the period
between actuation of the clean-agent FSS and reignition
because of ongoing propagation. Testing on the 2020 con-
tainer design indicated a maximum interval of 90 min for
the fire department to arrive on site, assess the situation,
consult documentation, and activate the water FSS. A
manual system would also present problems with certifi-
cation. An automatic system would meet certification
needs but would require a high-volume source of water to
be available at each container, which can present prob-
lems at remote sites.
Considering such issues, the newer codes and stan-
dards are having a significant influence on battery system
design for ESSs. The cells that use Li-iron phosphate (LFP)
chemistry, which is more resistant to TR, are becoming the
standard, despite their lower energy density. (The adop-
tion of LFP is also helped by lower cost.) Module manufac-
turers are taking extra steps to limit cell-to-cell
propagation of TR within their units. Beyond limiting
propagation, battery enclosures must be designed to avoid
or mitigate the buildup of potentially explosive gases
released during a TR event.
One of the challenges in UL9540A fire testing is the
interpretation of the performance requirements of the
module-, unit-, and installation-level tests. The intention
is to understand the performance under propagating TR
conditions, yet different chemistries respond differently
when in TR, and the test itself may introduce some vari-
ability in battery response compared to field conditions,
depending on the way in which TR is initiated.
For example, most NMC chemistries produce sparking
and often flames when TR is initiated using an external
heater, as in the UL 9450 A test. These flames can
(b) (c)
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 33
 consume vented flammable gases,
thus reducing a buildup of these
gases within the enclosure. In the
field, however, TR is more likely to be
initiated by an internal cell short,
where the cell’s energy rapidly reduc-
es as the temperature rises and vent-
ing occurs. This difference in cell
status at the point of venting raises
the question of whether there will
still be sparking and flames, and if
not, the resulting gas buildup will be
quite different. Furthermore, if flames
are suppressed by the discharge of a clean agent but TR is
allowed to propagate, there could be a considerable
increase in flammable gas concentration before a water
FSS is actuated. This situation should be seen in the UL
9540 A testing, but all possible conditions leading up to
such an event should be considered in a failure-modes-
and-effects analysis.
On the other hand, LFP cells are more stable at higher
temperatures and less likely to exhibit propagating TR.
When venting, they may not produce the sparking and
flaming normally seen in NMC products, and individual
cells can produce very high amounts of hydrogen and
other flammable gases per cell. This is particularly the case
as LFP cells are often manufactured with higher ampere-
hour capacities because of their lower risk. The designs of
enclosures to address this explosion hazard must base
their solutions on accurate gas generation data from test-
ing to the most current edition of UL 9540 A. The resulting
enclosure designs will incorporate systems to either
address any explosion pressure waves, or ideally, prevent
an explosion in the first place.
Managing vented gases has become one of the thorni-
est issues in Li-ion battery-enclosure design. The two
National Fire Protection Association documents that apply
here are NFPA 68, Standard on Explosion Protection by Defla-
gration Venting and NFPA 69, Standard on Explosion Preven-
tion Systems. These documents are called out in NFPA 855
as a binary choice, but the situation is not that simple.
Obviously, it would be preferable to prevent an explosion
in the first place rather than having an explosion occur
and managing it with deflagration venting. However, it
would be virtually impossible for an enclosure designer to
achieve compliance under NFPA 69 because that standard
does not allow for flammable gas accumulation above 25%
of the lower flammable limit (LFL) (60% of the LFL is also
allowed under certain conditions). The LFL for hydrogen is
4%, so NFPA 69 compliance would require that the con-
centration of that gas never exceed 1% (or 2.4%). Cell vent-
ing is quite violent, with the possibility of hundreds of
liters of gas (depending on the cell size) being released in
just a few seconds. With the need to design for the high-
est-possible energy density, there is little free space in
today’s battery enclosures, and it is inconceivable that the
34 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
Managing vented hydrogen concentration would not
rise in those few seconds to multiples
gases has become of the LFL.
Additional challenges are present-
one of the thorniest ed in gas detection due to the many
issues in Li-ion variables of a TR event. Examples of
these variables include the rate of gas
battery-enclosure evolution, variability of gas species,
mixing within the enclosure, cross
design. contamination of sensors, and fouling
due to the by-products of combustion.
With the seeming impossibility of
NFPA 69 compliance in preventing
explosions, does this mean that system designers should
ignore prevention and simply plan on managing explo-
sions? It is our opinion that this is a false choice and that
system designers should do what they can to meet the
intent of NFPA 69 by providing a means for rapid dilution
and exhaust of vented gas and, where space allows, to
also meet the letter of NFPA 68 by installing an appropri-
ate level of deflagration venting. An amendment to NFPA
855 addressed this issue, including correcting a gap for
ESS cabinets (where all the components are accessed from
exterior doors). The amendment provides for explosion-
control solutions in lieu of NFPA 68 or 69, in ESS cabinets
where fire testing demonstrates an absence of shrapnel,
projectiles, or pressure waves.
Beyond Li-ion
Battery technologies for ESS are not limited to Li-ion, par-
ticularly considering the future need for technologies
capable of providing low-cost storage for discharges of
longer than 4 h. Many new technologies have been pro-
posed, each promising lower cost at scale than Li-ion,
often with better cycle life or some other advantage.
Although there may indeed be a “Li-ion killer” out there, it
can be extremely difficult to determine which claims are
real and which are projected from early test results. It is
not unusual for a new technology to show very promising
results with small button cells in a lab, only to find that
the technology cannot be made to work in larger formats,
that manufacturing cannot be scaled up as expected, or
that unexpected failure modes show up, causing the new
batteries to fail prematurely.
Li-based batteries are far from immune from this sort of
hype. New Li-ion active materials are proposed with some
frequency, promising to be safer, cheaper, more energy
dense, or able to be recharged in scant minutes—and often
with all four features at the same time. Furthermore, there
is a new class of batteries using solid-state electrolyte and
metallic Li negatives, which promises all of these improve-
ments. Technologies of this type have been around since
the 1990s, although not in a form acceptable for mass-pro-
duced electric vehicles, and now enormous sums are being
spent by auto companies and venture capitalists to accel-
erate their development and industrialization. Although it
is difficult to go more than a few days
without reading about a new break-
Revised in 2020,
through in this field, the consensus of IEEE Std 1679 is
industry experts is that these new sol-
id-state Li batteries will not enter full- supported by a
scale production for electric vehicles
until approximately 2025.
growing series of
For these emerging technologies, subsidiary guides to
there is a need for a common frame-
work for manufacturers to character- its application for
ize their systems and for prospective
users to evaluate them. That frame-
different classes
work is provided by IEEE Std 1679, of batteries.
IEEE Recommended Practice for the
Characterization and Evaluation of
Energy Storage Technologies in Station-
ary Applications. Even though the primary focus is on bat-
tery technologies, the standard is applicable to other
storage media that provide a means for the reversible stor-
age of electrical energy; that is, the systems that receive
electrical energy and are able to release electrical energy
at a later time. The standard forms a foundation for the
objective evaluation of an energy storage technology by
providing a common basis for the expression of perfor-
mance characteristics, treatment of life-testing data, anal-
ysis of failure modes, and assessment of safety attributes.
Revised in 2020, IEEE Std 1679 is supported by a grow-
ing series of subsidiary guides to its application for differ-
ent classes of batteries. The published guides include IEEE
Std 1679.1 for Li-based batteries and IEEE Std 1679.2 for
sodium-beta batteries. Other guides are in preparation for
flow batteries and alkaline batteries. The following para-
graphs provide additional detail on each guide.
xx The initial release of IEEE Std 1679.1 was in 2017, and
there is a move underway to revise the document to
reflect the changing qualification environment and
emergence of newer technologies, especially solid-
state Li batteries.
xx The sodium-beta batteries covered by IEEE Std 1679.2-
2018 include high-temperature sodium-sulfur and
sodium-nickel chloride technologies, typically operat-
ing at around 270 °C. This group of technologies may
soon be expanded with the potential development of
new material capable of conducting sodium ions at
just above the melting point of sodium metal.
xx Flow batteries are the subject of the standards project
that is likely to be published next year as IEEE 1679.3.
This document will cover a broad range of redox and
hybrid flow chemistries. There is much interest in this
class of batteries, especially when required discharge
times start to exceed 4 h. It is thought that these lon-
ger discharge times will favor these technologies and
that they will finally emerge from the shadow of Li-ion.
xx The project for alkaline batteries is P1679.4, where “P”
denotes a not-yet-published standards project, and
has just commenced. This document will cover a
broad range of chemistries with non-
flowing alkaline electrolytes, many of
them zinc-based. The current list
includes zinc-air, nickel-zinc, nickel-
metal hydride, nickel-iron, and
rechargeable zinc-manganese dioxide.
Although nickel-cadmium chemistry
technically fits in this grouping, these
batteries are excluded because they
are already well characterized and
covered by other IEEE standards.
The IEEE Power & Energy Society
Energy Storage and Stationary Bat-
tery Committee remains on the look-
out for other classes of energy
storage technologies that could be
covered by a guide in the 1679 series. The base standard
covers emerging and alternative technologies, where
emerging technologies are those recently, or soon to be,
made available for sale under customary commercial
terms (e.g., defined scope of supply and warranted per-
formance). The alternative technologies are those that
are currently mature but less well known or frequently
deployed as traditional technologies. These definitions
keep the committee focused on technologies where pro-
spective users are genuinely in need of guidance in their
evaluation, rather than on those that have yet to emerge
from the laboratory.
(Inter)Connecting the Dots
It is one thing to develop safer and well-characterized
ESSs, but interconnection can remain a barrier to large-
scale deployment. The potential concerns with ESSs
depend on the level at which the interconnection is
made. In the distribution network, distributed energy
resources (DERs), including ESSs, are often set up to
export power to the grid, and there is a concern that if
this export continues when the grid has failed, numer-
ous problems could occur, including a potential safety
issue for line workers, who may be unaware that a feeder
is still energized. An additional concern relates to the
performance of grid services by aggregations of distribut-
ed systems, where the dispatch signal can shift rapidly
between charge and discharge, potentially creating dis-
turbances on a feeder. On the plus side, a grid-forming
ESS can be the lynchpin of a microgrid, providing resil-
ience to a facility that is normally grid-connected but can
disconnect as needed and function independently, pro-
vided that the microgrid has sufficient control and pro-
tection systems to meet the requirements of both
on- and off-grid operation. Having such a “sheddable
load” can be a benefit for the network operator, but the
transition to and from islanded mode must be achieved
without disruption.
At the transmission and subtransmission levels, the
concern is more over the loss of inertia in the bulk electric
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 35
 system as spinning generators are
replaced by renewable energy
resources and energy storage, all
of which connect to the network
through power electronics. The fast,
bidirectional response of energy stor-
age can contribute to the mitigation
of any stability issues, but such
responses by multiple resources must
be properly coordinated.
Standards are essential in address-
ing these issues. Interconnection
standards provide widely accepted,
consensus-based requirements and
best practices for connecting resourc-
es to power systems, and they help to
create consistency in policy across
multiple jurisdictions. IEEE has been writing interconnec-
tion standards for decades, starting with IEEE 929-1988,
which dealt exclusively with distribution-connected, resi-
dential photovoltaic systems, focused mainly on potential
power quality issues, and was only seven-pages long,
including an annex. Since then, the number, scope, and
length of standards have increased considerably and now
cover energy storage connected at the distribution or
transmission levels.
Large-scale energy storage will often be implemented
in the form of inverter-based resources (IBRs) intercon-
nected with transmission or subtransmission systems,
and there is a group of standards under development for
these interconnections. The first of these, IEEE P2800,
Draft Standard for Interconnection and Interoperability of
Inverter-Based Resources Interconnecting With Associated
Transmission Electric Power Systems, is expected to be
published later this year. (Note that the document is sub-
ject to change as it goes through the balloting process, so
statements in this article may not accurately reflect the
content of the final document.) IEEE P2800 defines a set
of minimum capabilities and performance requirements
for all IBRs connected to transmission or subtransmis-
sion systems. This new standard was crafted with energy
storage in mind and includes multiple examples, explan-
atory footnotes, and other materials illustrating the
Transmission Subtransmission
G IBR
IEEE P2800 (2021)
IEEE P2800.1
IEEE P2800.2
Figure 5. The IEEE standards relating to ESS interconnection.
36 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
Interconnection application of P2800 to transmission-
connected energy storage. This stan-
standards provide dard also addresses several common
applications of large-scale grid energy
widely accepted, storage, such as hybrid plants (e.g.,
consensus-based plants that include both photovoltaics
and energy storage) and fast frequen-
requirements and cy response.
IEEE Std 1547-2018, IEEE Standard for
best practices for Interconnection and Interoperability of Dis-
connecting tributed Energy Resources With Associated
Electric Power Systems Interfaces, is the
resources to power standard that applies at the distribu-
tion level, whether to individual sys-
systems. tems or aggregations that act as virtual
power plants. This document is a revi-
sion of IEEE 1547-2003. The 2003 ver-
sion of the standard was written in an environment in
which the aggregate power level of DERs on the system
was very low, such that they did not have a discernable
impact on transmission system behavior, and thus, the
2003 standard took the approach of tripping DERs offline
for many abnormal system conditions, leaving the system
to handle the condition. By the time the 2018 revision was
completed, the number of DERs on the system had
become sufficiently large that transmission impacts were
already evident, and thus, the 2018 revision contains
extensive requirements for capabilities that support the
bulk system, including ride-through requirements for
DERs to stay online over a much broader range of condi-
tions than before. Further addressing the unique charac-
teristics of distributed ESSs—with bidirectional flow of
both real and reactive power in the context of IEEE Std
1547—is the focus of a new standard, IEEE P1547.9, Draft
Guide to Using IEEE Std 1547 for Interconnection of Energy Stor-
age Distributed Energy Resources With Electric Power Systems,
expected to be published in 2022. An ESS connected at the
distribution level is referred to in this document as an
energy storage DER (ES DER).
The respective scopes of these interconnection stan-
dards and their subsidiary “dot-standards” are shown
graphically in Figure 5. The transmission-level require-
ments set by IEEE P2800 and the distribution-level re-
quirements set by IEEE 1547-2018
do have some significant differenc-
Distribution es. One key example of importance
to energy storage is the require-
ment for fast frequency response.
IBR ALL DER At the distribution level, IEEE 1547-
2018 does not explicitly require fast
IEEE 1547–2018 frequency response but does re-
IEEE 1547.1–2020 quire the capability to provide
IEEE P1547.2
frequency-droop response (a fre-
IEEE P1547.3
quency response proportional to
IEEE P1547.9
Tf). However, the shortest re-
sponse time allowed for this droop
function is 1 s, meaning that DERs
can react only during the recovery
This new standard
phase of a system-level frequency was crafted with
event, not the arresting phase. Hence,
this droop function is not specifically energy storage
referred to as fast frequency response in
in mind.
1547-2018. On the other hand, the
same 1-s period is the maximum re-
sponse time of fast frequency re-
sponse in IEEE Std P2800 and is typically within the
arresting period of a system-level frequency event. As a
result, this droop function could play a fast-frequency-re-
sponse role if the ES DER response time is parameterized
appropriately. The frequency-droop equation is given,
along with default parameter values and ranges over
which the parameter values may be adjusted. The fre-
quency droop applies for both high- and low-frequency
excursions. The utilization of frequency-droop capability
is by mutual agreement between the ES DER operator and
system operator.
IEEE 1547-2018 also allows ES DERs to perform “inertial
response” (a frequency response proportional to df/dt), but
it does not require this capability and it makes no recom-
mendations regarding its formulation or parameters. IEEE
P1547.9 provides more useful information in this regard,
but as a guide it cannot impose mandatory requirements.
Those requirements will have to wait for the next revision
of IEEE 1547.
At the transmission level, IEEE P2800 requires that
IBRs provide frequency droop for both positive and neg-
ative frequency deviations within the limitations of the
primary energy sources available to the IBR. To facilitate
this response, the draft standard defines two nameplate
ratings: continuous and short-term (surge) ratings. The
IBR plant can exceed the continuous rating, but not the
short-term rating, to provide frequency support. P2800
does allow the two ratings to be the same for those IBRs
that have no surge capability. This droop response is
referred to as a primary frequency response, and its speed
of response is limited only by the active-power ramp
rate of the IBR. In addition to that primary-response
droop requirement, IEEE P2800 also requires that all IBR
plants have fast-frequency-response capability for nega-
tive frequency excursions. This fast frequency response
is also formulated as a frequency droop, but as men-
tioned previously, the response time is required to be no
greater than 1 s. The standard gives the droop equation
for this fast-frequency-response function and provides
default parameter values and allowable ranges of
parameter adjustment. IEEE P2800 also allows inertial
response and provides details for its implementation in
an annex.
Looking to the Future
The ESS battery industry has come a long way from the
early “Wild West” days when each manufacturer made its
own decisions about safety. The rapid
development of codes and standards
has imposed a level of rigor in system
design, where safety is a critical compo-
nent at every step, rather than being
treated as an “add-on” after other require-
ments are met. The video record of
large-scale testing to UL 9540 A is an
important element that can provide a
strong sense of assurance to prospective users and the
AHJ alike.
That said, this is by no means the end of the story. New
lessons are being learned, and safety standards and codes
are being revised, even as this article is being written. New
requirements will be imposed on manufacturers, who will
have to be nimble in responding with necessary design
changes. This is a process that has no end.
The evolution of interconnection standards at both the
transmission and distribution levels will also provide a
pathway for increasing ESS deployments at all levels with-
in the grid. IEEE 1547 and the future IEEE 2800 will be piv-
otal in adapting the electricity system to one with more
widespread DERs, two-way power flows in the distribution
system, and lower inertia as rotating generators are sup-
planted by IBRs.
Acknowledgment
The release numbers for Pacific Northwest National Labo-
ratory and Sandia National Laboratories are PNNL-
SA-165876 and SAND2021-11431 O, respectively.
For Further Reading
“Test method brings clarity to insurance and fire mitiga-
tion professionals,” UL, UL 9540A. Accessed: Oct. 6, 2021.
[Online]. Available: https://www.ul.com/resources/
ul-9540a-test-method-brings-clarity-insurance-and-fire
-mitigation-professionals
“Fire fighter safety in battery energy storage system fires,”
NFPA, Quincy, MA. Accessed: Oct. 6, 2021. [Online]. Available:
https://www.nfpa.org/News-and-Research/Resources/Fire
-Protection-Research-Foundation/Current-projects/
Fire-Fighter-Safety-in-Battery-Energy-Storage-System-Fires
“Sandia National Laboratories/Pacific Northwest Nation-
al Laboratory Energy Storage Safety Collaborative,” U.S.
Dept. of Energy, Office of Electricity (DOE OE), Washington,
D.C. Accessed: Oct. 6, 2021. [Online]. Available: https://www
.sandia.gov/energystoragesafety-ssl/
Biographies
Jim McDowall (jim.mcdowall@saftamerica.com) is with
Saft America Inc, Jacksonville, Florida, 32221, USA.
Matthew Paiss (matthew.paiss@pnnl.org) is with Pacif-
ic Northwest National Laboratory, Richland, Washington,
99354, USA.
Michael Ropp (meropp@sandia.gov) is with Sandia
National Laboratories, Albuquerque, New Mexico,
87123, USA.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 37
 By Jeremy B. Twitchell and Alan S. Cooke
Emerging
Best Practices
for Modeling
Energy Storage
in Integrated
Resource Plans
An overview and a comparison.
NERGY STORAGE TECHNOLOGIES ARE A
E
unique class of energy resource, capable
of providing service across the electric
grid’s four functional levels: generation,
transmission, distribution, and customer
(Koohi-Fayegh and Rosen 2020). The rapid growth of
intermittent energy resources, driven by state and utili-
ty decarbonization goals, is creating a growing need for
energy storage to match unpredictable generation with
customer demand (Gür 2018). However, the primary
tool that many utilities use to identify grid invest-
ments—the integrated resource plan (IRP)—was not
designed to consider the unique characteristics of
energy storage.
In preparing an IRP, an electric utility forecasts the
demand of its customers and identifies the necessary
investments to meet that demand under various
Digital Object Identifier 10.1109/MELE.2021.3115557
Date of current version: 1 December 2021
38 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
future scenarios. The objective function of an IRP is to
create the lowest-cost portfolio of generation resourc-
es and demand-side measures (such as energy effi-
ciency and demand response) that will meet future
needs while minimizing the risk across the potential
futures studied.
Preparing an IRP is a complex endeavor that requires
multiple models and many simplifying assumptions.
There is a wide range of IRP modeling software devel-
oped for commercial use or by utilities for in-house
use, and those models perform a variety of functions.
At its core, however, an IRP requires the use of four
model types.
x Market forecast: Whether a utility operates within a
regional energy market or conducts bilateral trades
with neighboring utilities, the value of its generation
resources will affect and be affected by regional ener-
gy market conditions. By forecasting the performance
of regional energy markets, a utility can have an
informed estimate of the value of its existing and
2325-5897/21©2021IEEE

©SHUTTERSTOCK.COM/DUNCAN ANDISON
potential generation resources and a price forecast for
energy purchases.
x Load forecast: To identify the future demand of its cus-
tomers, utilities employ econometric models to proj-
ect customer usage based on economic, social, and
technological factors.
x Capacity expansion: To build out the utility’s system to
meet future demand, a capacity expansion model con-
siders various resource options (based on the cost and
performance assumptions that the utility builds into
the model) to identify the least-cost portfolio of genera-
tion and demand-side resources to meet that demand.
x Production cost: Where a capacity expansion model
identifies the cost of building a resource portfolio, a
production cost model simulates operation of that
portfolio for a representative year to verify that it can
meet reliability requirements and determine its annu-
al operational cost.
To account for different possible futures, the utility
will develop various scenarios for study. Example
scenarios that might be considered include the imposi-
tion of a tax on greenhouse gases (GHGs), the construc-
tion of a major new transmission line, or alternative cost
assumptions. Each scenario is run through the capacity
expansion and production cost models, resulting in mul-
tiple candidate portfolios that may then each be subject-
ed to further risk analysis by stochastically modeling
how they perform under varying levels of customer load,
fuel prices, and so on.
The objective of this process is to identify a preferred
portfolio: the combination of investments that minimizes
cost and risk across various possible futures. A preferred
portfolio does not represent a definitive plan for future
investments but, rather, a snapshot capturing the utility’s
best estimate of future needs based on current informa-
tion. Utilities typically prepare IRPs every two to three
years, and preferred portfolios may significantly differ
from one cycle to the next as circumstances change. To
guide the utility’s actions between IRP cycles, each plan
will usually contain an action plan outlining the steps
that will be taken to meet near-term needs. Figure 1 pres-
ents a diagram of this process.
For each scenario, the capacity expansion model
must build a portfolio that maintains a constant bal-
ance between load and generation. For each time
interval that the model considers, this means perform-
ing a complex calculation to find the optimal combina-
tion of demand-side resources, energy generation, and
market purchases to meet the demand forecast for
that interval. Because these calculations must be done
thousands of times for every scenario, utility planners
make several simplifying assumptions to keep the pro-
cess manageable.
Our previous work detailed how those simplifying
assumptions create barriers for energy storage, which can
be summarized in the following two types:
x Temporal barriers: To limit the calculations that they
must perform, capacity expansion and production
cost models operate at hourly granularity and use
reserve margins to represent ancillary service needs
(rather than modeling the optimal provision of those
needs). But these simplifications prevent models from
identifying the need for and value of flexible resources
like energy storage.
x Spatial barriers: To simplify the representation of
the utility’s system and maintain the focus on gen-
eration needs, IRP models focus only on the gener-
ation system. This prevents recognition of the
values that storage could provide to the transmis-
sion and distribution systems as well as recogni-
tion of the value that aggregated distributed energy
resources could provide to the generation system
(Cooke et al. 2019).
The previous work reviewed 21 IRPs from around the
United States to measure the degree to which utilities are
including energy storage in their planning processes. Two
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 39
 key findings from that work were that
utilities are relatively uncertain about
energy storage costs (as evidenced by
wide ranges in cost assumptions) and
that utilities that analyzed more ener-
gy storage services were more likely to
select storage in their preferred port-
folio (Cooke et al. 2019).
This article builds on that report
by identifying the specific ways in
which leading utilities are adapting
their planning processes and tools
to overcome these barriers and
more accurately account for the
many unique benefits of energy
storage. While the utilities repre-
sented in this report are not the
only ones that have taken such
steps, they were chosen based on
the transparency of how their practices were commu-
nicated in their IRPs and the measurable impacts of
the changes made, with the goal of capturing and
communicating the diversity of practices that have
been developed.
As Figure 1 illustrates, IRP processes are long and
complex. Because of this, there is significant optional-
ity in both the type and timing of improvements that
may be applied. By identifying a range of adaptive
Market Resource
Forecast Assumptions
Capacity Expansion
Portfolio Identification
Production Cost Modeling
Risk Assessment and Portfolio
Comparison
Preferred
Portfolio
Selection
Inputs Modeling
Figure 1. The IRP process.
40 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
The primary tool that tools and procedures that have
been deployed by utilities as well as
many utilities use to their relative complexity, this work
may be of use to utilities, regula-
identify grid tors, and other IRP stakeholders
investments—the interested in improving the repre-
sentation of energy storage in utili-
integrated resource ty resource planning.
plan—was not Northern Indiana Public Service
designed to consider Company
After Northern Indiana Public Service
the unique Company (NIPSCO) filed its 2016 IRP,
stakeholders questioned the lack of
characteristics of transparency in the utility’s cost
energy storage. assumptions for new resource
options, which were obtained from
proprietary sources and kept confi-
dential. In response, NIPSCO began
its 2018 IRP process by issuing a request for proposals
(RFP) to solicit current information on costs and per-
formance characteristics for various resources. NIPS-
CO received 90 proposals, which were aggregated and
supplemented with publicly available cost databases
to maintain confidentiality while allowing for public
disclosure.
NIPSCO initiated the process by studying cost assump-
tions for various resource options in seven publicly avail-
able reports. For a 4-h lithium-ion
battery, this study identified an
average assumed cost of US$2,110/
kW, but the assumptions ranged
Scenario
Identification from about US$1,300 to US$3,100/
Load
(Scenario 1, kW. Solar plus storage hybrid costs
Forecast
Scenario 2, were not identified in the initial
Scenario n)
study.
The RFP elicited nine bids for
standalone energy storage and 10 bids
for solar plus storage. Aggregat-
ed, the average bid for solar plus
storage projects was US$1,183/kW,
and the average bid for standalone
storage was US$1,349/kW. Using
these cost assumptions, NIPSCO
selected 92 MW of solar plus stor-
age by 2023 in its preferred portfo-
lio. (NIPSCO’s 2018 IRP is available
at https://www.nipsco.com/docs/
librariesprovider11/rates-and
-tariffs/irp/2018-nipsco-irp.pdf?
sfvrsn=16.)
Operating an RFP can be an
expensive and time-consuming
Outcome process for utilities. Since NIPSCO
only used the results of this RFP for
informational purposes, some of
the costs associated with conducting
in-depth review of proposals were
For each run,
likely avoided. The risk of this PLEXOS identifies
approach is that proposal develop-
ment is costly and time consuming, the cost savings for
and project developers may not
the portfolio that the
respond if they believe that the RFP
will only be used for informational flexible asset
purposes. Some utilities avoid that
issue by using less complex request creates by reducing
for information solicitations to generation from
obtain cost estimates.
higher-cost
Public Service Company of
New Mexico resources.
In its 2020 IRP, Public Service Compa-
ny of New Mexico (PNM) indicated
that, like most utilities, it assumes that limited amounts
of capacity may be procured through bilateral agree-
ments with other utilities as needed to maintain reliabili-
ty. But when resource adequacy challenges affected the
Western Interconnect in August 2020, and PNM experi-
enced an outage at a large thermal generator, the utility
was unable to procure replacement capacity through
bilateral trades. After reducing the amount of reserve
capacity that it assumed it could procure on the market,
PNM studied its energy needs and determined that
maintaining reliability would require the utility to
increase its planning reserve margin—the amount of
excess capacity that the utility procures to meet ancillary
services and serve as backup—from 13 to 18%.
Coupled with this analysis, PNM also conducted an
effective load-carrying capability (ELCC) analysis to quan-
tify the ability of energy storage technologies to contrib-
ute to meeting the increased planning reserve margin.
The ELCC is a well-established analytical approach that
measures how nondispatchable resources, like wind and
solar, and energy-limited resources, like energy storage,
contribute to maintaining reliability within a given ser-
vice territory, based on its load and generation mix.
PNM’s ELCC study determined that the first 300 MW
of storage on its system (PNM had no utility-scale stor-
age at the time) would have an ELCC of 95.7%, while the
next 150 MW would have an ELCC of 92.5%, and the fol-
lowing 200 MW would have an ELCC of 84.5%. ELCC val-
ues would fall more precipitously in subsequent
increments. With these analyses informing its modeling,
PNM’s preferred portfolio selected 300 MW of storage by
2022 and, depending on whether new gas generation is
authorized by regulators, an additional 223 to 507 MW by
2025. (PNM’s 2020 IRP is available at https://www
.pnmforwardtog ether.com/assets/uploads/PNM
-2020-IRP-FULL-PLAN-NEW-COVER.pdf.)
PNM’s reassessment of its planning reserve margin
needs and the ELCC analysis of storage’s ability to ful-
fill those needs are moderately complex and costly.
While the work requires sophisti-
cated analysis of a utility’s system,
it can be done in house by plan-
ning staff.
Puget Sound Energy
In response to direction from state
regulators to improve its modeling
of energy storage, Puget Sound
Energy (PSE) developed a new
“Operational Flexibility” analysis in
its 2017 IRP, which creates a hybrid
between traditional, hourly plan-
ning models and emerging, sub-
hourly planning models.
In PSE’s Operational Flexibility
framework, the utility conducts an
initial analysis using its in-house, hourly capacity
expansion model, the Portfolio Screening Model III
(PSM III). PSE then performed a production cost
model for each portfolio using PLEXOS, commercially
available planning software that is capable of sub-
hourly resource dispatch.
In PLEXOS, PSE runs each portfolio nine times, each
time adding a different flexible resource with well-
defined intrahour flexibility and constraints. In addition
to various configurations of batteries and pumped stor-
age hydro (PSH), these analyses considered gas-fired
options, such as a combined cycle turbine, a peaker, and a
reciprocating engine. For each run, PLEXOS identifies the
cost savings for the portfolio that the flexible asset cre-
ates by reducing generation from higher-cost resources.
Once PSE has determined the flexibility benefit for those
resources, it then reruns the PSM III capacity expansion
model with flexibility values added to applicable resource
options, allowing the model to reoptimize the portfolio
based on a more accurate accounting of each resource’s
system value.
The application of this framework had a clear
impact on PSE’s preferred portfolio. Most of the initial
scenarios that PSE studied did not identify a need for
energy storage until they added 50–75 MW by 2037,
but after the Operational Flexibility analysis was
applied, PSE’s preferred portfolio accelerated that need
to 50 MW of storage by 2023 and 75 MW by 2027. (PSE’s
2017 IRP is available at https://pse-irp.participate.online/
past-IRPs/2017.)
PSE’s hybrid approach significantly increased the repre-
sentation of storage and other flexible resources. While
the approach is not as thorough as others presented here,
in that it optimizes a given portfolio to be more flexible
rather than building a flexible portfolio from the ground
up, its modular nature allowed PSE to develop and imple-
ment it in a single planning cycle without disrupting the
utility’s existing IRP process. The approach is more
impactful from a cost perspective, however, as it requires
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 41
 the licensing and implementation of new commercial California Public Utilities Commission
modeling software. The IRP process in California is unique, as it is per-
In its 2021 IRP, the first completed since Washing- formed on a statewide basis, and state regulators—who
ton’s Clean Energy Transition Act mandated that elec- normally review and respond to utility IRPs—instead
tric utilities be carbon neutral by 2030 and achieve zero actively lead the process. Every two years, the California
emissions by 2045, PSE’s preferred portfolio includes Public Utilities Commission (CPUC) designates a 10-year
450 MW of distribution system-connected energy storage emissions target for the energy sector (subject to legis-
by 2045 and 375 MW of renewable plus storage hybrid latively established guidelines) and then prepares a Ref-
plants. (PSE’s 2021 IRP is available at https://oohpseirp erence System Plan (RSP) to identify the statewide
.blob.core.windows.net/media/Default/Reports/2021/ energy portfolio necessary to achieve the emissions
Final/IRP21_Chapter%20Book%20Compressed goal. Each of the state’s load-serving entities is then
_033021.pdf.) required to develop a conforming IRP that identifies its
obligations under the RSP and its plan for satisfy-
ing them.
TABLE 1. CAISO average hourly ancillary service Like other IRP processes, the RSP employs a
prices, 2020. capacity expansion model, Renewable Integration
Solutions (RESOLVE), and a production cost model,
Hour Regulation Regulation Spinning Re- Nonspinning
Ending Down (US$) Up (US$) serve (US$) Reserve (US$) the Strategic Energy and Risk Valuation Model
(SERVM). For its second biennial cycle in 2019–2020,
1 6.88 5.22 1.90 0.10 the CPUC applied two enhancements that improved
2 5.19 3.98 1.47 0.12 the RSP’s ability to value energy storage: modeling
improvements for SERVM and an energy storage
3 5.93 3.79 1.29 0.11
potential study to inform RESOLVE about how much
4 4.98 3.65 1.17 0.11 energy storage could feasibly be deployed in
5 5.30 3.75 1.33 0.11 the state.
In the first enhancement, the CPUC expanded the
6 4.96 4.64 2.27 0.12
services that SERVM allowed storage to provide from
7 6.31 9.15 5.12 0.13 regulation and frequency response to add spinning
8 10.00 7.52 2.94 0.12 and nonspinning reserves. As noted in our previous
9 14.18 6.61 1.52 0.14 work, as the number of services that storage pro-
vides in a model increases, resource plans are more
10 14.65 5.94 1.19 0.13
likely to identify cost-effective energy storage invest-
11 16.57 6.66 1.12 0.11 ments (Cooke et al. 2019). Table 1 presents the aver-
12 17.61 7.44 1.37 0.11 age hourly value of various ancillary services in the
California Independent System Operator (CAISO)
13 16.64 6.65 1.57 0.13
during 2020.
14 17.12 7.92 2.03 0.17 Ancillary service markets are shallow relative to
15 15.85 8.78 2.69 0.51 capacity and energy markets (Chen et al. 2017).
Therefore, while the new services reflected in
16 11.75 8.93 3.82 1.06
SERVM model (spinning and nonspinning reserve)
17 10.48 12.04 7.50 2.66 are generally lower in value than the regulation ser-
18 7.91 22.74 18.64 11.09 vices that were already in the model, their inclusion
increases revenue opportunities and market size for
19 5.93 35.06 31.39 23.66
energy storage. Storage that would not have found
20 5.42 29.90 26.37 16.98 room in regulation markets could instead provide
21 5.64 13.52 9.96 4.10 spinning and nonspinning reserves, which for sever-
al hours each day are higher-value services.
22 4.73 8.74 5.09 1.43
In the second enhancement, the CPUC commis-
23 6.00 6.41 3.07 0.39 sioned a study to identify the upper bound of energy
24 6.70 5.87 2.03 0.12 storage that could help meet the state’s future needs.
When energy storage is used as a capacity resource
  Hour’s most valuable service
for reducing peak needs, as peaks become flatter,
  Hour’s second-most valuable service incremental storage investments must have longer
  Hour’s third-most valuable service durations to reduce the remaining peak. Figure 2
illustrates this point, using example load data for an
  Hour’s least-valuable service
evening in spring 2019 from CAISO.

42 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

 After a certain point, 4-h storage The IRP process in undertaking. The external study, on
devices can only meet a portion of the other hand, was a relatively low-
the peak demand, which diminishes California is unique, cost and simple means of informing
the value of the capacity that they planning assumptions regarding the
provide. The study that the CPUC as it is performed on a potential for energy storage on the
commissioned used an ELCC statewide basis, and California grid.
approach similar to PNM’s. But rather
than identify the inflection points at state regulators—who Portland General Electric
which ELCC values would decline, In its last two IRP cycles, Portland Gen-
the objective of the CPUC’s study was
normally review and eral Electric (PGE) has taken major
to quantify the total amount of 4-h respond to utility steps to improve its modeling of ener-
energy storage that could contribute gy storage technologies. Because of the
to California’s capacity needs in a IRPs—instead actively evolutionary approach the utility has
high-renewables future before those employed in this process, it has devel-
diminishing returns would render
lead the process. oped multiple distinct practices that
additional investments uneconomic. will be discussed here.
The study’s authors determined that,
because of the aggressive projected buildout of solar and PGE’s 2016 IRP: Net Cost Analysis
wind generation in California in the next decade, the Recognizing that its traditional, hourly capacity expan-
state could add more than 10 GW of 4-h energy storage sion model was not capturing the granular benefits of
before reaching that point (Carden and Wintermantel flexible resources, PGE developed a secondary model to
2019). This practice of identifying the technical potential study those benefits in its 2016 IRP. The Resource Optimi-
of a resource to inform planning assumptions has been a zation Model (ROM) was designed to dispatch resources
common practice for energy efficiency measures in many at both subhourly and real-time resolution while
states for years. accounting for the system’s ancillary service needs, there-
The institution of these changes increased the amount by capturing the intrahour benefits of energy storage and
of energy storage selected in the RSP from 2,104 MW by other flexible options.
2030 in the 2018 plan to 11,384 MW by 2030 in the 2020 Once those benefits were identified, PGE projected
plan. (The CPUC’s RSP is available at http://docs.cpuc. them for an entire year to generate an annual “opera-
ca.gov/PublishedDocs/Published/G000/M331/ tional value” for each studied asset. PGE looked at six
K772/331772681.PDF.) different battery configurations with various capacity
Relative to other measures presented in this article, and duration characteristics and, for comparison pur-
the two actions taken by the CPUC fall at different points poses, also studied the operational value of a 25-MW
on the spectrum. Developing and refining granular combustion turbine (CT). Where PGE’s traditional model-
modeling software represents a costly and complex ing tools indicated that the cost of a 25-MW, 4-h battery

25,000
24,000
23,000
CAISO Load (MW)

22,000

21,000
20,000
Initial Load Shape
19,000 Load Shape With Incremental 192 MW of 0.5–1-h Storage
Load Shape With Incremental 1,069 MW of 1.5–2-h Storage
18,000
Load Shape With Incremental 735 MW of 2.5–3-h Storage
17,000 Load Shape With Incremental 763 MW of 3.5–4-h Storage
16,000
15,000
18 18:30 19 19:30 20 20:30 21 21:30 22 22:30 23 23:30 24
Time of Day

Figure 2. Increased duration requirements for incremental energy storage assets to reduce peak generation needs in CAISO. (Source: Data
from the International Energy Agency: https://www.iea.org/data-and-statistics/charts/the-california-duck-curve.)

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 43

 would be about US$5 million more
per year than a 25-MW CT, the ROM
analysis found that the higher oper-
ational value of the battery reduced
the differential to about US$2 mil-
lion per year. Though the CT
remained the lower-cost option, the
net cost study indicated the price
point at which energy storage assets
would be cost competitive in future
planning. (PGE’s 2016 IRP is available
at https://downloads.ctfassets.net/41
6ywc1laqmd/1y737MdERELNLNyWA
W8bw2/3f150507210f0fba46276de38c
0afdfa/2016-irp.pdf.)
The primary benefit of the net cost
approach is that it can be implement-
ed with minimal disruption. By iden-
tifying the subhourly benefits of a
resource that other IRP models do not
capture, those costs can be levelized
and deducted from the resource’s assumed costs or
applied as an added value to its modeled output. In this
way, the net cost approach complements, rather than dis-
places, a utility’s existing IRP models.
A secondary benefit of the net cost approach is that it
can be implemented at minimal cost. PGE’s in-house
development of the ROM tool was relatively complex
and costly, but multiple entities have since developed
free energy storage modeling tools for use by utilities
and other parties. The Pacific Northwest National Labo-
ratory’s Energy Storage Evaluation Tool (https://eset
.pnnl.gov) and the Electric Power Research Institute’s
Storage Value Estimation Tool (www.storagevet.com) are
two such examples.
PGE’s 2019 IRP: Valuing Flexibility
PGE’s 2019 IRP established what is an apparent first for a
modern IRP by selecting a preferred portfolio that includ-
ed both new PSH and long-duration battery storage. This
IRP was also among the first in the United States to
select dispatchable, behind-the-meter (BTM) energy stor-
age in its preferred portfolio. These landmark selections
were the result of three significant changes that impact-
ed the way energy storage was analyzed: preventing the
model from selecting new GHG-emitting generators,
enhancing and integrating the ROM model into the core
planning process, and including BTM energy storage as a
resource option.
The no-GHG constraint was a policy decision that
PGE made based on state policies and customer prefer-
ences. But the model’s selection of 200 MW of PSH is
indicative of future investments in long-duration ener-
gy storage that will be necessary to achieve decarbon-
ization goals, and similar scenarios may be useful for
other utilities to analyze long-term system needs as a
44 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
When energy storage growing number of states, munici-
palities and customers adopt clean
is used as a capacity energy goals.
Whereas PGE used the ROM as an
resource for external model for one-off analyses to
reducing peak inform other IRP models in the 2016
IRP, the 2019 plan fully integrated the
needs, as peaks ROM into the process, using it to cal-
culate a flexibility value—a more
become flatter, detailed version of the operational
incremental storage value used in the 2016 IRP—for all
resource options. The ROM calculated
investments must the flexibility value by modeling the
system a week at a time with three
have longer levels of temporal resolution: hourly
durations to reduce unit commitment on a day-ahead
basis, followed by 15-min unit com-
the remaining peak. mitment on an hour-ahead basis, and
concluding with 15-min real-time dis-
patch. By stepping through different
levels of granularity, with unit com-
mitments made in each period preserved in subsequent
periods, PGE was able to drill down into its real-time ancil-
lary service needs with high resolution and select the opti-
mal resources to meet them.
But PGE also looked at its flexibility needs at the hourly
level and found high-demand periods on winter mornings
in which its resources would be stretched thin over several
hours and the lack of flexibility could create reliability chal-
lenges. This finding created a higher flexibility value for lon-
ger-duration batteries, which drove PGE’s selection of
37 MW of 6-h batteries in the preferred portfolio.
The final change in PGE’s 2019 IRP was the inclusion of
utility-controlled BTM storage as a resource option. The
change was prompted by an external study that the utili-
ty commissioned to forecast customer adoption of vari-
ous distributed energy resources. The study concluded
that customer uptake of energy storage technologies
would be limited by economics and technological famil-
iarity, but that if the utility were to offer incentives in
exchange for operational control of customer-sited stor-
age, then customers would be more likely to invest in the
technology (Navigant 2019). The study also quantified the
benefit streams of the devices to the utility and the cus-
tomer, allowing PGE to calculate an incentive level to cus-
tomers based on the value to the utility and then model
that incentive as an asset cost in the IRP. With a full
accounting of both the costs and benefits of this resource,
the system expansion model identified a cost-effective
level of 4 MW of dispatchable BTM energy storage by
2025. (PGE’s 2019 IRP is available at https://downloads.
ctfassets.net/416ywc1laqmd/6KTPcOKFlLvXpf18xKNseh/
271b9b966c913703a5126b2e7bbbc37a/2019-Integrated
-Resource-Plan.pdf.)
Because it utilizes a fully integrated, granular model
for determining the flexibility value of all resource
options (as opposed to just a subset of resources) and The rankings presented in Figure 4 are subjective
then weaves those values into its modeling, PGE’s 2019 based on our experience. Practices are positioned in
IRP represents a uniquely thorough, holistic approach to relative terms. In general, practices that involve an
modeling energy storage and
other flexible resources. However,
because it required the internal
development and integration of PNM:
CPUC: PGE (2019):
NIPSCO: Reserve PGE (2019):
new software and steps into the Storage External
Enhanced Study and GHG
IRP process, it is also among the Potential BTM Storage
Cost Inputs ELCC Constraint
Study Study
most complex and costly practices Analysis
studied. PGE’s gradual approach to
this challenge—developing and
Scenario
refining it over multiple planning Identification
Market Resource Load
cycles—provides a template for (Scenario 1,
Forecast Assumptions Forecast
managing those complexities. Scenario 2,
Scenario n)

Summary and Conclusion

A common theme among IRPs
reviewed for this work is that, as PGE Net Cost
Capacity Expansion
(2016): External
the electric grid transitions to a PSE:
Flexibility Model
cleaner, more variable resource Portfolio Identification Operational
mix, energy storage will be increas- Flexibility
PGE (2019): Full
Analysis
ingly necessary. Though traditional Integration of
Production Cost Modeling
IRP modeling tools lack the tempo- Flexibility Model
ral and spatial granularity to identi-
Risk Assessment and Portfolio
fy cost-effective opportunities to CPUC: Additional Comparison
deploy energy storage, the nine Storage Values
innovative practices identified in in SERVM
this work demonstrate that there Preferred
are many ways to overcome those Portfolio
Selection
barriers. The diversity of these prac-
tices creates multiple access points
for a utility, regulator, or other
Inputs Modeling Outcome
stakeholder seeking to improve the
treatment of energy storage in an
Figure 3. IRP entry points for best practices.
IRP. Figure 3 illustrates the points
where these practices improve the
IRP process.
These practices also vary in
terms of complexity and cost, from PSE
Increasing Cost

(Operational Flexibility)
simple measures that can be
implemented quickly at minimal PGE
cost to highly complex measures CPUC (ROM Enhancement)
(Potential Study)
that may require significant cost
and take years to fully realize. This PGE
variation ensures that incremental (BTM Study) PGE
steps are available to any utility, (Net Cost)
NIPSCO
regardless of the sophistication of
(RFP for CPUC
its current IRP model. These exam- Costs) PNM (SERVM)
ple IRPs also demonstrate that utili- (ELCC)
PGE Net Cost
ties may take an evolutionary
(No-GHG (Generic)
approach that develops and refines Scenario)
practices over multiple planning
Increasing Complexity
cycles. Figure 4 presents these best
practices in terms of relative com-
plexity and cost. Figure 4. The relative cost and complexity of identified best practices.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 45

 external study to inform assump-
tions (such as the CPUC’s storage
potential study and PGE’s BTM
study) are assumed to be less com-
plex but more costly. Practices that
involve the internal development or
refinement of modeling software
are assumed to be relatively more
complex, but, because they use
staff resources and are the intellec-
tual property of the utility, they are
assumed to be relatively less expen-
sive than licensing a commercial
software tool. And while PGE’s in-
house development of the ROM tool
to support its energy storage mod-
eling was a time-consuming and
likely costly endeavor, the cost and
complexity for a generic utility to
perform a net cost analysis is
assumed to be much lower due
to the existence of free modeling
tools. And while PGE’s scenario
prohibiting new sources of GHGs is
the least complex and costly practice, absent other
modeling enhancements to overcome the temporal
and spatial barriers of capacity expansion models, it
would only be capable of identifying long-duration
storage needs.
The innovative practices identified in this work repre-
sent a significant investment from the entities that
implement them, but those innovations have generally
not dispersed to the broader utility industry. Most of the
IRPs presented in this work are from utilities and other
entities located in the Western United States, where the
steady evolution of planning practices has resulted in
modeling approaches that can be transparently described
in IRP documentation, creating a clear link between the
analytical process and plan outcomes. While several
other utilities in the Southeast and Midwest have includ-
ed energy storage in recent IRP preferred portfolios, their
IRPs generally do not describe the analytical steps taken
to reach those conclusions.
Where planning outcomes are not clearly supported
by the modeling process, the IRP process breaks down.
IRPs are a primary source of evidence use by utilities to
justify investments to regulators and recover those
investments in rates. While the utilities cited in this arti-
cle are providing extensive narratives about how they
conducted the IRP and how they reached their conclu-
sions, other IRPs reviewed by the authors selected energy
storage and other resources without any clear narrative
46 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
The innovative explaining how or why those out-
comes were reached. That lack of
practices identified supporting evidence complicates the
regulatory review process and expos-
in this work es utilities to the risk of regulatory
represent a rejection of investment decisions.
Taking steps to incorporate the
significant best emerging practices identified in
this article can help utilities, regula-
investment from the tors, and stakeholders improve the
entities that transparency of an IRP and facilitate
informed review and consideration of
implement them, but planning outcomes.
those innovations Acknowledgment
have generally not We acknowledge and thank the Ener-
gy Storage Program within the U.S.
dispersed to the Department of Energy–Office of Elec-
tricity, under the direction of Dr. Imre
broader utility Gyuk, for funding this work.
industry.
For Further Reading
K. Carden and N. Wintermantel, “Energy
storage capacity value on the CAISO sys-
tem,” Astrape Consulting Rep., 2019. [Online]. Available: https://
www.astrape.com/?ddownload=9137
H. Chen, S. Baker, S. Benner, A. Berner, and J. Liu, “PJM inte-
grates energy storage: Their technologies and wholesale
products,” IEEE Power Energy Mag., vol. 15, no. 5, pp. 59–67,
Sept.–Oct. 2017. doi: 10.1109/MPE.2017.2708861.
A. L. Cooke, J. B. Twitchell, and R. S. O’Neil, Energy Storage in
Integrated Resource Plans. Richland, WA: Pacific Northwest
National Laboratory, 2019. [Online]. Available: http://energy
storage.pnnl.gov/pdf/PNNL-28627.pdf.
T. M. Gür, “Review of electrical energy storage technolo-
gies, materials, and systems: Challenges and prospects for
large-scale grid storage,” Energy Environ. Sci., vol. 11, no. 10,
pp. 2696–2767, 2018. doi: 10.1039/C8EE01419A.
S. Koohi-Fayegh and M. A. Rosen, “A review of energy stor-
age types, applications and recent developments,” J. Energy
Storage, vol. 27, Feb. 2020. doi: 10.1016/j.est.2019.101047.
“Distributed resource and flexible load study: Integrat-
ed resource planning system-level report,” Portland Gen-
eral Electric, Navigant, PGE IRP, 2019. [Online]. Available:
https://downloads.ctfassets.net/416ywc1laqmd/6KTPcOKF
lLvXpf18xKNseh/271b9b966c913703a5126b2e7bbbc37a/
2019-Integrated-Resource-Plan.pdf
Biographies
Jeremy B. Twitchell (jeremy.twitchell@pnnl.gov) is with the
Pacific Northwest National Laboratory, Richland, Washington,
99352, USA.
Alan S. Cooke (alan.cooke@pnnl.gov) is with the Pacific
Northwest National Laboratory, Richland, Washington,
99352, USA.

By Mohammed Ben-Idris, Michael Brown,
Matthew Egan, Zhenyu Huang, and Joydeep Mitra
Business models for utility-scale
shared energy storage systems
and customer participation.
Utility-Scale Shared
Energy Storage
UE TO CLIMATE CHANGE, SUPPLY SCARCITY,
D
and society’s desire to expand access to
electricity and improve energy-system resil-
ience, there has been an increasing demand
to invest in and use renewable energy
sources (RESs) that are environmentally friendly, efficient,
sustainable, and affordable. This has diversified and decen-
tralized energy sources and increased their penetration.
However, the variability and intermittency of RESs has
introduced new challenges to power system operators,
such as the requirement for high ramping rates when solar
Digital Object Identifier 10.1109/MELE.2021.3115558
Date of current version: 1 December 2021
2325-5897/21©2021IEEE
©SHUTTERSTOCK.COM/STOCKWERK-FOTODESIGN
energy declines in the afternoon, known colloquially as the
duck curve, voltage and frequency fluctuations, and
reduced overall system inertia. In response to these chal-
lenges, energy storage systems (ESSs) (devices such as bat-
teries, energy management, and energy conditioning) have
become crucial components to the reliable and stable oper-
ation of modern power systems. ESSs can tackle the afore-
mentioned challenges to seamlessly integrate RESs into
the power grid. They can help system operators to smooth
the output power of RESs and provide grid services, such as
voltage and frequency regulation, peak load shaving, phase
balancing, energy arbitrage, expansion deferral, and so on.
ESS sizes range from large-scale storage systems, such
as pumped-storage hydropower (PSH), to small batteries
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 47
 installed in home appliances. Each type and size of ESS has
specific uses, operation requirements, and challenges. In
this article, we discuss the utility-scale shared energy stor-
age (USES) battery systems that are operated by a distribu-
tion utility and available for use by customers under
various business models—not to be confused with com-
munity energy storage systems (CESs). USES systems pro-
vide fractional storage to customers under various
metering, pricing, and billing approaches. Generally, there
are three levels of battery-type ESSs that have been used at
the distribution-system level: at-home, CES, and USES.
ESSs at home can be either directly behind the meter
(BTM) or at the point of connection for home appliances.
CESs are usually installed at the secondary side of distribu-
tion transformers and used by groups of customers (e.g.,
homes on a residential feeder and commercial buildings).
USES systems are much larger than CESs and installed at
the substation. They are usually owned by utilities or inves-
tors. Figure 1 shows the applications of battery ESSs at dif-
ferent levels of an example distribution system.
USES systems provide many advantages over CES and
BTM storage systems: USES systems are more flexible and
less expensive than other types of ESSs for comparable
sizes, can offer grid service at scale, improve grid reliability
Commercial
Substation
USES
EMS
Load
Figure 1. The applications of (battery) ESSs at different levels: USES, CES, and local residential and commercial ESSs.
48 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
and resilience at scale, and reduce fire and other hazards in
homes. Photovoltaic (PV) owners and other community
members can purchase a fractional portion of a USES sys-
tem to help manage their premise load and electricity
bill. This fractional portion operates as a virtual storage
block allocated for customer use. However, there are many
challenges in the development of USES systems. These
challenges include developing technology and business
processes that effectively manage energy flows for both cus-
tomer use and grid services use cases. The principle among
the challenges is to determine the optimal charging/discharg-
ing schedules for the virtual storage blocks and the battery
as a whole while considering both the customer rate struc-
ture and the economic value derived from grid services.
History
As mentioned previously, a variety of ESSs have been used
at different levels. The choice of a given energy storage
device depends on the needed application, although here
we are mostly interested in the grid applications that pro-
vide grid services, and at the customer side, for demand
flexibility. Large storage systems, including pumped
hydrostorage and compressed-air energy storage, have
been developed and deployed for decades. However, until
EMS
Residential
CES
EMS EMS
PV
recently, battery applications have been limited to small-
scale usages such as uninterruptible power supplies, back-
up in some applications, bridging with backup generators,
at-home applications, auxiliary services at power plants,
and so forth. Large-scale applications, such as grid- and
utility-scale batteries, have been mostly in lab develop-
ment up until the past decade.
The last 10 years have yielded an exponential increase
in the manufacturing and installation of large-scale bat-
teries. The main technologies in use have been sodium-
sulfur, vanadium-redox, lithium-ion (Li-ion), and lead-acid
batteries. For example, a 200-MW/800-MWh, vanadium-
redox flow battery-storage project is under construction in
Dalian, China, and will be the world’s largest battery-stor-
age facility when completed. Li-ion batteries are widely
used and applied in a significant majority of grid-scale,
battery-storage projects. The world’s largest solar (850 MW)
and battery system (531 MW/2,125 MWh) has been
approved for construction in the desert north of Las Vegas,
Nevada—there is some opposition due to its environmen-
tal impact as well as it being an eyesore to local residents.
In large-scale energy storage deployment, the state of Cali-
fornia was the first to reach gigawatt-scale deployment
and achieved its 2020 1,325-MW energy-storage goal
ahead of schedule. California further projected 55,000 MW
of new storage by 2045. Arizona Public Service is planning
to install 850 MW of storage by 2025. Southern California
Edition (195 MW) and Pacific Gas and Electric (567.5 MW)
have received approvals from the California Public Utilities
Commission to build storage facilities.
Publicly available data from the U.S. Energy Informa-
tion Administration regarding battery storage only go as
far back as 2003, when the largest—and only—battery
facility in the United States had a nameplate rating of 27 MW
in the battery ESSs in Alaska. Within the past 10 years,
large-scale batteries have been installed at an exponential
rate, as depicted in Figure 2. They went from being
installed in only two facilities with nameplate capacities
under 30 MW in 2010, to having more than 1.5 GW of
nameplate capacity at more than 230 locations in only 10
years. This was driven largely by the falling costs of batter-
ies, which went from prohibitively expensive at more than
US$1,000 per kWh, to under US$200 over the same decade.
Therefore, from an economic perspective, it only makes
more sense to install storage systems in a fashion similar
to the precipitous decline in price for PV solar panels.
Financial and Operational Challenges
Although USES systems offer control and management
flexibility for system operators and equity and affordabili-
ty for customers, many challenges remain for the wide-
spread application and utilization of USES systems. The
biggest challenges relate to the management of energy
storage, distribution, degradation, and ownership, which
can result in greater financial benefits or in burdens for
the utility and customers. One of the biggest hurdles in
executing USES systems is developing and employing
business models that not only manage bills and agree-
ments between utilities and users but also allow utilities
and communities to benefit from energy charging/dis-
charging to provide a stable and resilient power supply
and necessary grid services. Although seemingly complex
by design, only a few studies have made efforts to design a
business model for shared battery storage.
Besides the economic considerations, challenges with
USES also arise from an operational perspective. USES
offers multiple potential applications, including firming
renewable generation, supporting frequency and voltage
profiles, mitigating transient stability, providing local black-
start capabilities, enabling energy arbitrage, and enhancing
demand response. These applications are achieved at vari-
ous time frames and need to be carefully managed so that
the state-of-charge (SoC) of the battery system is main-
tained at an optimal level to enable these applications
when needed. These can often create a conflict for the
stored energy among the intended applications. For exam-
ple, the battery should have an adequate SoC level when
an outage occurs and black start is called for, charging and
discharging rooms should be reserved so that the battery
can absorb or release energy in the range of renewable
variations the battery intends to mitigate, and the inverter
control should be designed to manage multiple functions
and coordinate competing objectives. Although none of
these needs are mutually exclusive, they can be close to
zero sum as prioritizing one need may come at the cost of
another. Ultimately, all of these applications translate into
financial terms, as either gained benefits or incurred costs
in the pursuit of an optimal solution. These costs and ben-
efits should be properly assessed and represented in the
business model for optimal financial performance and,
more importantly, provide incentives for USES to offer
these applications for any and all interested customers.
Environmental Challenges
Operational challenges are relatively well understood with-
in the industry, but there is no product without its
U.S. Battery Capacity
Nameplate Capacity (MW)
1,800
1,600
1,400
1,200
1,000
800
600
400
200
0
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
Year
Figure 2. The reported battery USES growth in the United States in
the past 20 years. The information was gathered from publicly avail-
able U.S. Energy Information Administration data obtained from utility
companies.
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 49
 externalities. USES systems do have
them, and they are worth noting. The
two main forms of battery storage are
Li-ion and lead acid. Lead-acid batter-
ies have very clear disadvantages as
they are from one of the most hazard-
ous metals known to humanity. Li-ion
batteries, on the other hand, do not
have the same health drawbacks that
come from lead-acid batteries, or from
their disposal. Their difficulties lie on
the other end of the supply chain.
Li-ion batteries, the main form of
battery storage for USES, can create
significant environmental damage
during the procurement and extrac-
tion of lithium. Current lithium mining techniques involve
both strip and brine mining. Strip mining involves the
stripping of a mountain or landscape to find the minerals
underneath. This can be very difficult for plants and ani-
mals, especially the endangered species that may rely on
elements in the area. Brine mining can have just as signif-
icant an impact as strip mining, although not in the same
way. It is a system wherein aquifers are deliberately
drained and allowed to evaporate so that the minerals
within it can be extracted. Brine mining can be damaging
to water-scarce areas, such as the American Southwest,
which may need that water for agriculture, conservation,
or domestic uses.
This is seen especially in the Thacker Pass project, which
has faced heavy backlash from environmentalists over con-
cerns for local species, which live on top of what is one of
the largest proven reserves of lithium in the United States.
This can cause cognitive dissonance for environmentally
minded utilities, which may rightly see the benefits that
USES can give to improving the viability of renewables, but
also do not wish to see more immediate harm come to eco-
logically significant areas or to endangered species.
None of this is to say that large-scale batteries would
do more harm than good. There are clear and observable
benefits to using more storage, which has helped make
things such as solar and wind generation far more viable
components of the grid. It is only to say that policy should
be directed in a considered manner for the pros and cons
of each energy storage method.
Services From USES Systems and Value
Proposition
USES systems can provide grid services and various
forms of support at the distribution level. These include
energy services (energy time shift and supply capacity),
ancillary services (regulation, voltage support, and so on),
transmission and distribution (T&D) infrastructure servic-
es (T&D upgrade deferral, transmission congestion relief,
and so forth), and customer energy-management services
(power quality, demand charge management, and so on).
50 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
We discuss some of these services in
An ESS can be the following sections.
utilized to store
Reliability Services: Outage
energy during off- Mitigation
Service interruptions can be caused by
peak hours and then many events, such as failures of gener-
discharges at peak ating units, or T&D components such
as transformers, lines, and feeders. In
demand periods for such situations, a USES can effectively
peak shaving or load support customer loads when a partial
or complete loss of power from the
following. source utility takes place. Sometimes,
due to the capacity constraint, it might
not be possible for a USES to mitigate
the outage completely. However, for
such an event, it can shorten the interruption duration or
reduce the number of interrupted customers. Blocks can
also be appropriately allocated to customers based on load
criticality to provide a framework that supports reliability-
differentiated services. USES also creates a better opportuni-
ty to recover from said outages, as unlike traditional turbine
systems, they are able to come back online very rapidly,
allowing for local microgrid creation, which can supply criti-
cal loads while the grid at-large comes back online.
Energy Arbitrage
An ESS can be utilized to store energy during off-peak hours
and then discharges at peak demand periods for peak shav-
ing or load following. This helps to reduce the generation
cost and postpone the need for peaking units. It is also prof-
itable for the energy storage owners as they can take advan-
tage of the energy price difference. In competitive markets,
locational marginal prices (LMPs) indicate the value of ener-
gy at different locations and points of time. To benefit from
this application, the USES can be charged at off-peak hours
or with less-expensive energy, and discharged during peak
hours or when LMPs are high. The locations with large vari-
ability in LMPs will generally provide higher margins.
Frequency Regulation
Energy storage is also well suited for frequency regula-
tion. Frequency regulation helps to maintain grid fre-
quency within specified limits and comply with the Real
Power Balancing Control Performance (BAL001) and Dis-
turbance Control Performance (BAL002) Standards of the
North American Electric Reliability Corporation (NERC) or
any other relevant governing agency. The system’s fre-
quency tends to deviate from the specified value as gen-
eration and load imbalances occur during normal
operation. Traditionally, several generator actions are
taken to restore and maintain frequency within the nor-
mal operating range. These include primary, secondary,
and tertiary frequency control and response times, which
range from a few seconds to several minutes. A USES can
help in such a situation, either by acting as an extra load
to lower the frequency when supply exceeds demand, or
to discharge when demand exceeds supply. Suitably
interfaced, it can be twice as effective in regulation com-
pared to conventional fossil fuel generators such as coal
units and combustion turbines.
The aforementioned services have often been used to
quantify the “payback” of USES installations, although in
many cases, the payback has been lower than the invest-
ment cost. Perhaps the most lucrative service has been
that of frequency regulation. PJM Interconnect has been a
leading driver in this space, initially remunerating gener-
ous amounts in capacity and performance payments.
However, other markets have been slow to follow, and the
payouts have declined as the proliferation has increased.
Yet it is perhaps unfair to judge the value of USES installa-
tions based on payback alone. There are several other ser-
vices that are very beneficial to the system, but the impact
on payback is hard to determine.
Other Services and Benefits
USES can also be used for various other applications that
are crucial to the power grid in several ways, but they
might be difficult to demonstrate in terms of monetary
benefits. These applications can be classified according
to the customers being served, e.g., residential, commer-
cial, or industrial customers. The following are a few of
these applications:
xx Voltage-flicker mitigation: Voltage flicker is often a prob-
lem that needs to be addressed in power distribution
systems. It is harmful for customers of all load pro-
files. Voltage flicker might damage electrical devices
ranging from the most common household applianc-
es household to industrial-scale equipment. It might
also lead to spoilage from semifinished industrial
products. Thankfully, USES can help stabilize the volt-
age by charging or discharging within a very short
time, thus protecting the customer’s equipment.
xx Power-factor improvement: This problem is mostly
applicable to industrial customers that use a signifi-
cant amount of reactive power for their daily opera-
tion. In most of the cases, they are charged with a
penalty from the utility serving them if the power fac-
tor goes below a prespecified limit. These customers
can greatly benefit from the presence of USES, which
can act similarly to supercapacitors by providing reac-
tive power while also providing more general uses.
xx T&D upgrade deferral: USES can also be used for the
deferral of T&D system upgrades. It can help in delay-
ing investments that would otherwise be necessary to
maintain the T&D capacity in accordance with the
load demand. For example, the purchase of a new
transformer with a high capacity may be avoided by
using a USES instead.
xx Voltage-regulator lifetime extension: The addition of dis-
tributed generation, e.g., PV or wind power to a feeder
in the distribution system, might decrease the lifetime
of a voltage regulator due to a rise in the number of tap
changes with an increase in voltage fluctuation. If USES
is suitably used to smooth variability, then it can reduce
the number of tap changes of the regulator by reducing
the voltage fluctuation, thus extending its lifetime.
Emission reduction: Climate change and global warming
xx
are matters of concern, especially with regards to emis-
sions from fossil fuel plants. Thermal power plants
across the world are among the largest consumers of
fossil fuels. The generation of renewable resources and
their integration into the grid reduces emissions and
the burning of conventional fuels. USES can accumu-
late clean energy at off-peak hours, e.g., when wind
power generation is high at night, while load demand
is low, and then discharges it at peak hours, thus
improving its effectiveness at reducing emissions.
Business Models
USES systems have gained significant momentum to pro-
vide opportunities for owners of RESs to lease portions of
their storage instead of buying individual ESSs. In this con-
text, new multiuse business models, which utilize battery
storage for different grid services and improve power supply
reliability and resilience for local communities, are needed
to increase economic benefits for USES owners (e.g., utilities)
and customers with BTM distributed energy resources
(DERs). In theory, USES can provide many benefits for indi-
vidual households and the community. For example, black-
outs and shortages due to extreme weather conditions can
be mitigated with an ESS. However, before this scheme can
be implemented, utilities, investors, and community mem-
bers need to know how to assess its costs and benefits. As
mentioned previously, developing and employing business
models for USES systems can be a challenge due to compli-
cated interactions between utilities, system operators, and
BTM participants. Also, USES can provide several benefits to
both utilities and investors (e.g., BTM participants) with dif-
ferent perspectives, which can introduce conflicting objec-
tives. For utilities, the benefits can include increasing the
hosting capacities of distribution systems to electric vehicles
and high penetration of RESs, deferral of distribution system
expansion, and fewer communication requirements as
communicating with one point at the substation level
requires less communication than communicating with
many ESSs at homes or BTM devices.
Grid operators can utilize USES systems to provide grid
services at scale, including frequency regulation, voltage
control, and demand-side management. For investors,
USES systems can provide an affordable and safe energy
storage through leasing blocks of the shared energy stor-
age; this is extremely important for the occupants of mul-
tifamily buildings and tenants. Such an offering would also
benefit property owners through a reduced maintenance
burden. This fee may be similar in nature to already-exist-
ing renewable energy fees, wherein ratepayers who wish to
ensure that they are funding reliable technologies will pay
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 51
 a premium. For this case, it may be
necessary to have a certain amount of
power reserved to shorten an outage
or to allow for a flatter rate overall.
This may appeal to risk-averse rate-
payers who would rather not have to
worry about how costs vary with the
time of day or how temporary faults
may affect the grid around them.
USES systems can provide reve-
nue for both the utility and custom-
ers through energy arbitrage, which
can be managed through business
models and agreements. Under the
existing cost structure, ESSs have
been deployed to only provide grid
services that utilize less than 50% of
the total battery lifetime capacity,
which cannot justify economic bene-
fits to owners and investors. To fully utilize USES systems,
their use and benefits can be managed through storage-
leasing agreements.
Utilities need to adapt business, financial, and process
models for a mass-market, energy-storage-as-a-service
(ESaaS) business model for utility-owned USES systems.
One ESaaS model candidate that is under investigation by
NV Energy—a vertically integrated utility in Nevada—
includes a fixed fee for a managed services offering in
which customers pay the utility a calculated monthly fee
per virtual storage block on a subscription basis. This partic-
ular model uses ESaaS as one component in an overall Inte-
grated Services Bundle (ISB) in which the customer takes
service under a particular rate, and the utility uses targeted
demand management to achieve both customer bill sav-
ings and the provision of grid services. The EsaaS customer
payments contribute to recovering the installed equipment
cost plus the allowed return on capital, and ongoing main-
tenance. This approach reduces the utility rate-base
requirement and minimizes or eliminates cross subsidiza-
tion from other customers.
There may also be a justification, both in public rela-
tions and in the aforementioned savings to upgrading sys-
tems, to simply include energy storage in the general cost
of electricity. This may be best implemented in remote
areas that suffer from low voltages at peak loads. This
would help to lower the stress or sag on transmission
lines. This can be done in a way similar to how utilities
may already require customers to install capacitors as
methods of power-factor correction. This would, of course,
require a significant amount of analysis on the cost of
power line upgrades versus the cost of battery installation,
especially if it would require hiring outside firms for the
management of said batteries, but it is only likely to
become more viable as the cost of batteries drops.
The utilities would need to work with local govern-
ment and regulatory stakeholders to reach a consensus on
52 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
Building and the business models that are compati-
ble with the utility’s regulatory com-
implementing pact, and identify the changes (if
necessary) to existing franchise agree-
business models for ments. Working with local govern-
USES requires the ments, business models for USES
systems can be designed to accelerate
consideration of the clean energy transition and car-
bon-reduction goals among the mass-
stakeholder market customers that do not have
perspectives, access to the financial or technical
resources required to dramatically
including customer, expand energy efficiency, load flexibil-
ity, and site-renewable energy.
regulatory, and
utility perspectives. Customer Willingness to
Participate in USES
Building and implementing business
models for USES requires the consid-
eration of stakeholder perspectives, including customer,
regulatory, and utility perspectives. The following sections
further review the ESaaS and ISB approach under devel-
opment at NV Energy with partners at University of
Nevada, Reno (UNR), University of Nevada, Las Vegas
(UNLV), and Evolution Networks under contract to the U.S.
Department of Energy. It includes a discussion of product
development criteria as well as survey data collected
from customers of NV Energy in Nevada to determine
their willingness to participate in grid services and lease
blocks of USES.
Integrated Service Bundle
One of the approaches to consider when developing ESaaS
products is to include them in a tailored ISB that provides
positive perceived use value to customers through a bun-
dle of financial and environmental benefits.
NV Energy started research into an ISB developed for
single family residential customers with rooftop solar tai-
lored to its regulatory environment that includes:
xxDynamic rate: a dispatchable, time-varying rate treat-
ment with a demand component that provides pric-
ing signals intended to align consumption patterns
with electric grid costs and support nonwire solutions.
xxESaaS: a low-cost, virtual storage offering to further
enhance the customer’s ability to derive savings from
the rate treatment.
xxGrid service payments: the opportunity to earn money
by allowing their DER devices to participate in the grid
service set.
xxHome-automation gateway: a premise hardware gate-
way that integrates into the customer’s DERs to enable
the aforementioned offerings via intelligent agent soft-
ware while also providing added convenience features
in the form of a home-automation gateway, which the
customer can use for a wide variety of both energy and
nonenergy-related automation functions.
Customer Criteria
The overall value proposition (of the
70 BTM Option
ISB), considering the tradeoffs asso-
ciated with making and receiving Remote Option
60 Price Not Specified
payments and receiving enabling
technology (e.g., automation devic-
50
es), must generate positive, custom- Price: US$15
Price
er-perceived value that includes Specified Per Month
40
financial savings. Each element of Price: US$25
the service bundle should be Per Month
30
designed such that the service 1) is Price Not
Specified
easily understood by customers, 2)
has demonstrable perceived use 20
value, 3) has sufficient customer
demand, 4) is congruent with cus- 10
tomer willingness to pay, 5) does
not present an unacceptable level of 0
risk, and 6) is fair and transparent.
Building USES projects includes Figure 3. Customer willingness to participate in ESaaS for BTM and remote options (i.e., shared
customer market research to deter- energy storage option).
mine whether there is sufficient
market demand among customers for ESaaS and grid ser- other DERs are willing to participate in and utilize USES
vices. The service elements must also meet a series of reg- systems has revealed the following.
ulatory requirements. Across a range of adoption and program scenarios that
included options for utility-owned and managed BTM stor-
Regulatory Criteria age and remote “virtual” storage, the survey validates that
From a regulatory perspective, service elements must sat- monthly energy bill reduction is the top driver of DER adop-
isfy a series of requirements, which include 1) compliance tion. For the BTM configuration, when price and savings
with existing regulations; 2) responsiveness to local gov- were not specified, 58% of survey respondents said they
ernment policy directions; 3) no returns that exceed an “definitely” or “probably” would participate in an ESaaS
allowed rate of return; 4) no unreasonable costs for rate- offering. That level dropped to 39% when price and savings
payers, yet comprising a positive benefit-to-cost ratio; were specified and there was no statistically significant dif-
5) no unreasonable risks for ratepayers; 6) usefulness and ference across either the price or savings levels tested.
helpfulness to fulfill resource adequacy and operational For the remote configuration, when price and savings
needs; and 6) consistency with the overall resource plan. were not specified, 23% of survey respondents said they
Electric utilities must prove that these requirements are “definitely” or “probably” would participate. That level
met via a series of regulatory assessments and filings. increased to 32% when price was specified at US$25 per
month and further increased to 40% when price was
Utility Criteria specified at US$15 per month. There was no statistically
Utility criteria focus on operational and economic consid-
erations. At a minimum, the introduced system must not
generate unacceptable physical, cyber, or information
security risks. It is also important that they comply with
NERC’s Critical Infrastructure Protection reliability stan-
dards and do not generate unacceptable levels of project
cost recovery risk. It is essential that they provide a clear 38%
and observable benefit to grid management and/or quality Low Income
and integrate effectively with both existing and future
62%
enterprise systems. Multifamily

Customer Market Research

As mentioned previously, executing USES projects should
include customer market research, which can be done
through conducting surveys and analyzing customers’
responses. A survey conducted by NV Energy to determine Figure 4. The willingness of multifamily and low-income customers
whether customer and owners of BTM PV systems and to participate in ESaaS.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 53

 significant difference across the saving levels tested. The
results of customer willingness to participate in ESaaS
for both BTM and remote (virtual) options are summa-
rized in Figure 3.
An examination of the demographic data revealed
large differences in the percentage of customers who “def-
initely” or “probably” would participate in a shared energy
storage offering (see Figure 4). Low-income and multifami-
ly customers (38 and 62%, respectively) expressed the
highest levels of interest; however, the sample size of the
solar-adopting multifamily group was low (n=8) and mer-
its further research to validate statistically significant dif-
ferences from the single-family group. The research
suggests that remote virtual storage, although quite
attractive for single-family customers with high-energy
bills, is probably more attractive for low-income and mul-
tifamily customers; hence, NV Energy expanded their ini-
tial development efforts to include an ISB approach for
multi-family and low-income customers without existing
rooftop solar or storage, which is less technologically com-
plex and provides a simplified billing based approach for
the ESaaS component in the form of a storage share, simi-
lar to a program developed at Sacramento Municipal Utili-
ty District.
Among storage adopters, 74% of respondents said they
“definitely” or “probably” would participate in a grid ser-
vice offering. The respondents expressed that a perfor-
mance-based compensation methodology was the top
choice, along with a discounted electricity rate. The perfor-
mance-based methodology was explained as a “payment
per unit of energy drawn from the battery.”
Conclusion
The development and deployment of grid- and utility-
scale battery ESSs have been exponentially increasing
over the last decade as a result of several factors, including
advancements in fabrication materials for batteries and
the need to smooth the variability of RESs. Specifically, the
applications of USES systems have been on the rise
because they provide flexibility and controllability for both
grid operators and customers. Power grid operators can
use USES systems in several grid services (e.g., frequency
and voltage regulation, resource adequacy, energy arbi-
trage, and the deferral of system upgrades) with less com-
munication and control requirements than CESs and BTM
storage. Also, customers can lease blocks of USES systems
instead of installing local batteries that can be costly,
inflexible, and introduce hazard concerns. In addition to
these benefits, USES systems provide equity, reliability,
and resiliency to local communities. However, USES sys-
tems still face several challenges, including building busi-
ness models that can accurately capture charging/
discharging costs and benefits, designing control systems
capable of optimizing and prioritizing specific grid servic-
es with customer needs, and developing tools to improve
the round-trip efficiency of USES systems and reduce
54 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
losses in the grid due to charging/discharging actions initi-
ated by customers.
Acknowledgment
Part of the findings presented in this article are based
upon work supported by the U.S. Department of Energy’s
Office of Energy Efficiency and Renewable Energy (EERE)
under the Solar Energy Technologies Office Award Number
DE-EE0009022 and the U.S. National Science Foundation
(NSF) under Grant NSF 1953648. Neither the United States
Government nor any agency thereof, nor any of their
employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accura-
cy, completeness, or usefulness of any information, appa-
ratus, product, or process disclosed, or represents that its
use would not infringe privately owned rights. The corre-
sponding author is Mohammed Ben-Idris.
For Further Reading
“Form EIA-860 detailed data with previous form data (EIA-
860A/860B),” The U.S. Energy Information Administration,
Washington, D.C. Accessed: Sept. 15, 2021. [Online]. Available:
https://www.eia.gov/electricity/data/eia860/
Y. Tian, A. Bera, M. Benidris, and J. Mitra, “Stacked revenue
and technical benefits of a grid-connected energy storage
system,” IEEE Trans. Ind. Appl., vol. 54, no. 4, pp. 3034–3043, July/
Aug. 2018. doi: 10.1109/TIA.2018.2825303.
P. Denholm et al., “The value of energy storage for grid
applications,” National Renewable Energy Laboratory, Golden,
CO, Tech. Rep. NREL/TP-6A20-58465, May 2013. [Online]. Avail-
able: https://www.nrel.gov/docs/fy13osti/58465.pdf
“Community batteries: A cost/benefit analysis,” Australian
National University, Canberra. Accessed: Aug. 15, 2021.
[Online]. Available: https://arena.gov.au/assets/2020/08/
community-batteries-cost-benefit-analysis.pdf
California Public Utilities Commission. https://www.cpuc
.ca.gov/news-and-updates/all-news/cpuc-orders-clean
-energy-procurement-to-ensure-electric-grid-reliability
(accessed Oct. 10, 2021).
P. Pettingill, “Ensuring RA in future high VG scenarios – A
view from CA,” in Proc. ESIG Spring Workshop, Apr. 10, 2020.
Accessed Oct. 10, 2021. [Online]. Available: https://www.esig
.energy/event/2020-spring-technical-workshop/
Biographies
Mohammed Ben-Idris (mbenidris@unr.edu) is with the
Department of Electrical and Biomedical Engineering, Uni-
versity of Nevada, Reno, 89557, USA.
Michael Brown (michael.brown@nvenergy.com) is with
NV Energy, Las Vegas, Nevada, 89146, USA.
Matthew Egan (matthewegan@nevada.unr.edu) is with
the Department of Electrical and Biomedical Engineering,
University of Nevada, Reno, 89557, USA.
Zhenyu Huang (zhenyu.huang@pnnl.gov) is with Pacif-
ic Northwest National Laboratory, Richland, Washington,
99352, USA.
Joydeep Mitra (mitraj@msu.edu) is with the Depart-
ment of Electrical and Computer Engineering, Michigan
State University, East Lansing, Michigan, 48824, USA.

 By Jonas Kristiansen Nøland
Vacuum transport as a clean alternative to short-haul flights.
HE FUTURE TRANSFORMATION OF TRANS-
T
portation is set to fundamentally shape
our modern civilization in reducing the
global energy consumption from travel
and the time needed to move. In the
future climate vision of the European Union (EU), 90%
of travel-related emissions will be omitted by 2050. To
achieve this optimistic goal, one of the EU’s strategical-
ly proposed solutions is the vacuum train (vactrain)
concept, known as the hyperloop transportation
Evolving Toward a
Scalable Hyperloop
Technology
Digital Object Identifier 10.1109/MELE.2021.3115542
Date of current version: 1 December 2021
2325-5897/21©2021IEEE
©SHUTTERSTOCK.COM/VITAMIN444
system (HTS) in 2013 (European Commission 2020). The
hyperloop is currently viewed as the fastest way to
cross Earth’s surface (Nøland 2021). It employs fully
electric propulsion and, thus, it is seen as a clean
option. The optimal routes that are relevant for imple-
mentation are found in traffic-intense intercity
regions, which would generate sufficient throughput
to pay back its infrastructure investment. Another
driver is that European countries (e.g., France and Ger-
many) have already proposed banning short-haul
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 55
 domestic flights where travel by
rail is an option. Such policies
could accelerate the push for new
hyperloop projects as a promising
alternative to rail. In this article, a
technical path for affordable hyper-
loop development and implemen-
tation is presented.
T h e l o n g - f o rg o t t e n i d e a o f
va ctrains was already patented
decades ago (Goddard 1950). Follow-
ing the closure of the Swissmetro
vactrain project in 2009, the key
idea of a low-pressure tube was
rebranded as the hyperloop in an
open source design paper in 2013
(Musk 2013). In its core principle,
passenger or freight capsules are
propelled inside an airless vacuum
tube at nearly the speed of sound.
Hyperloop is still at the early stage
of development, as no fully operational systems exist,
and many engineering challenges are left open. In Fig-
ure 1, a conceptual HTS is depicted, including magnetic
lift and drag for suspension, air losses, and forward
thrust. Here, the capsules generate their own propul-
sion, driven by linear motors and powered by batteries
on board. As a result of recent marketing hypes about
the technology, not enough explanation of its function-
ality has been provided, which is the goal of this article.
Technical Drivers of Hyperloop Development
The key benefits that are promoting the development and
implementation of hyperloop are as follows. By removing
the air from the tube, the advantage is dramatically reduc-
ing the air losses and the associated drag, thus enabling
ultra-efficient transportation means. The feasible cruising
Lift
Motion
Thrust
Weight
Figure 1. A conceptual illustration of the HTS configured with magnetic levitation (maglev) and electric propulsion. A free-body diagram of arrows
indicating the different forces acting on the capsule is depicted.
56 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
The optimal routes speeds are rivaling the fastest air-
planes, with an energy consumption
that are relevant for per passenger (PAX) per kilometer [or
revenue passenger-kilometer (RPK)]
implementation are that are very close or inferior to those
found in traffic- of trains and comparable to those of
electric cars. For longer routes, the
intense intercity energy consumption per RPK can be
reduced further. This is because a
regions, which would larger share of the tube will be utilized
generate sufficient for the cruising speed.
In many ways, hyperloop can be
throughput to pay seen as a symbiosis between the
aviation and rail industries. It is a
back its subsonic train that experiences high-
infrastructure altitude atmospheric pressure levels
and must cope with safety certifica-
investment. tion standards similar to or stricter
than those of airplanes. The impacts
of hyperloop on humans are a signifi-
cant reduction in delay and travel
times, and that transits can happen much closer to the
big city centers. Benefits will be further explained in the
following sections.
Recreating the Outer Atmosphere
The main advantage of the hyperloop concept is the
opportunity to achieve a high-altitude low-pressure envi-
ronment at the surface of Earth. In reality, airplanes have
to consume a lot of energy to reach the 10-km altitude,
where the pressure level is 75% lower. Approximately
27 kWh of potential energy is needed to lift every ton of
mass from sea level to 10 km above the ground. In addi-
tion, one would also have to displace a distance where
drag forces must be met as well. These energy needs
come on top of the kinetic energy needed for an airplane
to accelerate to 900 km/h, which constitutes about 9 kWh
Lift
Air Loss
Magnetic Drag
Weight
per ton of mass. The kinetically
stored energy is more difficult to
regenerate on an airplane (i.e., regen-
erative soaring), which is in a non-
constrained environment, opposite to
the hyperloop.
To keep a medium vacuum on
Earth, vacuum pumps must be oper-
ated along the tube. The concept can
be made economically viable and
more energy-efficient than airplanes,
given that the number of capsules
utilizing the tube environment gener-
ates enough throughput in the sys-
tem. A steady passenger flow and
easy embarking must be kept as long
as the tube is operated. The high amount of energy need-
ed to depressurize the tube favors operating the tube at
low pressure continuously for day and night and as long
as possible to justify the energy consumption needed to
depressurize (Nøland 2021). One attractive solution would
be to allow freight transport during the night and more
passenger-related travel during the day, to maximize your
revenue as much as possible. This synergy might be
important, as the freight market is usually much bigger
than passenger needs. A passenger-only transportation
system would rarely be profitable and might need govern-
mental subsidies to operate.
Energy is also needed to compensate for the leakage of
air into the tube (i.e., concrete’s air leak). This is another
good reason for the importance of keeping the tubes oper-
ated continuously. By lowering the tube pressure as much
as possible, the higher are the gains in reduced air loss.
However, the desire for a low-pres-
sure environment must be weighed
against its safety concerns. It is also Outer
worth noting that a hard vacuum 100 km Space 10 km
would be too expensive. On the
30
extreme end, the hyperloop con-
Hyperloop is still at portion, i.e., 0.1% of the air resistance
experienced at atmospheric pressure
the early stage of (i.e., 1 atm). In addition, the ideal gas
law states that any heat accumulated
development, as no inside the tube will cause a tempera-
fully operational ture rise that reduces the air loss fur-
ther. Figure 2 illustrates how the
systems exist, and power consumption due to air resis-
tance is reduced by lowering the pres-
many engineering sure as well as increasing the
challenges are temperature. The power consumption
is presented for two cruising speeds,
left open. 1,000 km/h and 1,235 km/h, respec-
tively. Compared to sea level and aero-
space altitudes, the aerodynamic
power consumption in the hyperloop
environment seems insignificant in Figure 2.
Figure 2 is realistic given that the drag scales linearly
with the air density. The drag equation predicts that
the total drag experienced by a body submerged in air
is proportional to the dynamic pressure [0.5 ) t ) v 2
(Nøland 2021)], the frontal surface area of the body (A),
and the drag coefficient (C d). The premise for a low air
resistance relies on the realism of a low drag coefficient
inside the tube. This only holds if the tube’s volume
surrounding the capsule can be assumed infinitely
large, relative to the vehicle. Normally, the walls of the
tube are much closer to the capsule. Inevitably, we are
dealing with an internal aerodynamic problem, where a
violation of the so-called Kantrowitz limit could poten-
tially increase the drag, and thus, deuterate the bene-
fits of the hyperloop concept. Let us have a look at how
this concern can be mitigated.
Increasing Altitude 0 km
0 °C
Aviation
Hyperloop

20 °C
cept could be designed with a 99.9% 25 40 °C
lower pressure than at sea level, i.e., 60 °C
Drag Power (MW/m2)

m /h 80 °C
as low as it would recreate the k 100 °C
20
, 2 35
1
atmosphere experienced closer to n ic– 0 °C
o
the outer space, implying a massive ns 20 °C
15 Tra 40 °C
reduction in drag-related energy 60 °C
10 80 °C
consumption. 100 °C
Sea Level

km/h
5 ic– 1,000
A Dramatic Reduction of S ubson
the Air Resistance
0
The density of air is approximately 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
proportional to the pressure where Pressure (atm)
it exists. As a result, the air density
inside the tube is only 1/1000 of Figure 2. The drag power is plotted per frontal surface area of a capsule moving at subsonic
atmospheric air with a 99.9% lower (1,000 km/h) and transonic (1,235 km/h), respectively. The projections are calculated from the
drag equation (Fd = 0.5 C d tAv 2), where the drag power (or air loss) is the drag force (Fd) times the
pressure. The aerodynamic drag speed (assuming a unitary drag coefficient, C d = 1). Ideal gas law (p = RTt) is used to obtain the
will then be reduced with the same air density (t) as a function of both pressure (p) and temperature (T ), where R = 287.05 J/kgK.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 57

 The hyperloop capsule resembles a
subsonic or near-sonic wind tunnel,
where choked flows and shock waves
at the tail easily occur. The flow
around the capsule will accelerate and
create choked flows if the blockage
ratio is high (i.e., the capsule’s cross-
sectional area is large relative to the
tube). In these cases, the sides of the
capsule are too close to the tube. Elon
Musk’s white paper from 2013 pro-
posed that a huge compressor in the
front of the capsule could release the
flow and prevent the pressure build-
up by mitigating the twisted flow
around the pod. However, the required
compression ratios were several times
higher than the current jet engines. Consequently, it is per-
ceived that this solution rather complicates the pod design.
It is possible to deal with the increased drag by
approaching other options instead. For a cruising speed of
1080 km/h, it has been identified a nonlinear relationship
between the drag coefficient and the blockage ratio
(Nøland 2021). This ratio describes the ratio of the cross-
section frontal area of the capsule to the inner tube area. A
25% ratio means that the diameter of the capsule is about
half of the inner diameter of the tube. Hence, more
depressurization energy is needed to fill the tube, but it
can significantly reduce the drag, as more space is avail-
able around the capsule. The violation of the so-called
90 1,
50
250 k
It might also be Kantrowitz limit then reduces. A feasi-
ble drag coefficient could be as low as
more fruitful to deal a factor of two times the unity drag
coefficient (Nøland 2021). As a result,
with any additional the penalty applied to the estimated
extra drag by drag power in Figure 2 (i.e., with unity
drag factor) is significantly lower than
strengthening the reduced air resistance due to the
reduction in the pressure level.
the propulsion As seen earlier, a better path for-
system instead of ward might be optimizing the aerody-
namic performance of the capsule or
adding compressors increasing the tube volume. It might
also be more fruitful to deal with any
on board. additional extra drag by strengthening
the propulsion system instead of add-
ing compressors on board. Still, it is an
open design question to minimize global energy use.
Feasibility of Shorter Travel Times
Another major benefit of the hyperloop is its dramatic
reduction in travel time. However, due to the acceleration
needed, the travel time depends strongly on the length of
the route where it is built. There is always a needed trajec-
tory time and length before the capsule reaches the
desired cruising speed. As a result, the travel time does not
improve significantly beyond 1000 km/h for routes in the
range between 250 km and 500 km. This effect can be
observed in Figure 3, where the travel time is plotted as a
function of the cruising speed.
Figure 4 considers the impact
of the acceleration, with a base
thrust of 0.1 G (assumed in Figure

500

1,0

0 3) and with a final cruising speed

km
00

of 1200 km/h. As the velocity pro-

km
m

km

75
Ch file develops, the propulsion power
ica
go needed to accelerate increases lin-
–B
60 os
Travel Time (min)

ton early with speed (v). It also scales

Ma
drid
– P linearly with the normalized
Pun a
45 e–C ris g-force, which is the ratio of the
hen
nai acceleration (a) to the gravitational
Berlin–Kra
30 San Fran kow constant (g = 9.81 m/s 2). The accel-
cisco–Lo
Los Ange
s Angele
s eration power (Pacc) can be calculat-
les–Las V ed from the following expression:
egas
15 Amsterdam
–Eindoven
Pacc = mg a g k v,
a
0
0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 2,200 2,400
where the mass of the capsule (m)
Cruising Speed (km/h)
could be taken as 1,000 kg to identify
Figure 3. The relationship between travel time (in minutes) and cruising speed (km/h) is shown the acceleration power needed per
for different journey distances, 250 km, 500 km, 1,000 km, and 1,500 km, respectively. Some ton. A similar relationship applies
example routes are also depicted. The impact of the acceleration trajectory is taken into account,
with a mean acceleration of 0.1 G and a deacceleration of 0.3 G. The G-force values were chosen to the deacceleration process as
based on the report by the Transportation Economics & Management Systems report “Great Lakes well. The maximum power needed
Hyperloop Feasibility Study” (https://www.glhyperloopoutreach.com/feasibility-study). The cruising will strongly influence the
time is calculated by dividing the cruising length by the cruising speed, and the acceleration and
braking times are calculated by dividing the cruising speed by the mean acceleration or deacceler- dimensioning of the capsule’s pro-
ation, respectively. pulsion system (either on board or

58 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

external). One solution is to restrict the acceleration when
the vehicle has reached a certain speed level and then
speed up toward the desired cruising velocity.
Figure 4 illustrates the strong nonlinear relationship
between the maximum acceleration power and the
distance needed to accelerate. First, it can be observed
that a reduction of 25–50% has less effect on the distance
needed, while restricting further has a much stronger
impact. The maximum power rating should, therefore, be
chosen carefully. One would be less interested in

1,200
% 75%
0 25%
10
50%
1,000

800
Speed (km/h)

600

400

200

0
0.12
Constant
Thrust 100%
0.1
75
%
0.08
G-Force

0.06 50%

Constant
0.04 Power
25%
0.02

0
350

0%
300 10
Acceleration Power (kW/tonne)

75%
250
245.2
kW/tonne
200
50%
150 163.5
kW/tonne
100 25%
81.75
50 kW/tonne

0
0 20 40 60 80 100 120 140 160
Distance (km)

Figure 4. The impact of restricting the acceleration power by constraining the g-force during the acceleration trajectory distance for a capsule
with 0.1 G in initial acceleration and a final cruising speed of 1,200 km/h. The calculation is based on Newton’s second law of motion
(RF = ma), the accumulation of velocity (v = # a dt) , as well as the accumulation of distance (s = # v dt). The acceleration power is the product of
the thrust and the instantaneous speed (Pacc = RFv), and the G-force acceleration is normalized with the gravitational constant (g = 9.81 m/s 2).

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 59

 overdimensioning the power reser-
voir on board the capsule if the
acceleration distance does not get
significantly reduced. It would be a
worthless investment. Moreover, it
has been shown that high accelera-
tion performance increases the cap-
sule’s overall mass as well as the
energy consumption per passenger-
kilometer (Tudor and Paolone 2019).
Figure  4 suggests that partial restric-
tions in the acceleration do not have
detrimental effects on the overall per-
formance. For instance, at 1 g-force,
the acceleration length needed to
reach 1,200 km/h would be only
5.6 km, 1/10 of what is needed at 0.1 G. However, it would
be a less comfortable experience for the passengers. In the
extreme case, where maximum acceleration power is
reduced by 75%, the speed profile is restricted when the
capsule has reached 300 km/h. Due to the limited acceler-
ation from 300 km/h to 1,200 km/h, the trajectory profile
toward cruising gets significantly longer. Even though it
downrates the propulsion system, the tube trajectory
needed to accelerate gets about two and a half times lon-
ger. Increasing the capsule’s trajectory traveled at lower
speeds also influences the feasible travel time.
An Environmentally Friendly Alternative to
Conventional Rail
Finally, another environmental advantage of hyperloop
over railways is that the tube can be built above the
ground on pylons with prefabricated sections. The sup-
porting pillars might be installed for every 30 m on aver-
age and tall enough to cause reduced disruption of
farmland. It also emits less noise from its operation since
it can contain its sound inside the tube. Moreover, the tube
infrastructure is a perfect synergy for solar harvesting that
lowers the occupation of the environment. One of the
major challenges with solar energy is the occupation of
land areas. It can be estimated that a solar roof on top of
the tube can make the hyperloop net energy positive.
Based on the California climate, a two-way tube of 4.25-m
width can provide an electric power capacity of 510 kW
per km of the tube at peak solar activity (assuming
120 W/m 2) or an annual average of 75.5 kW/km (i.e.,
16.78 W/m 2).
The Hyperloop Propulsion Concepts
This section takes the technical drivers on board to
explore the realization of different hyperloop propulsion
solutions. Even though everything has not yet been set-
tled, most major companies now agree to use linear elec-
tric motors for propulsion and employ maglev for
suspension and guidance. Moreover, the typical target for
the pressure levels inside the vacuum chamber is in the
60 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
Thus, the self- range between 1/1000 of the atmospher-
ic pressure (1 atm) to the pressure levels
propelled pods are where aviation has been certified to fly,
i.e., 1/10 of sea level.
economically the
best alternative, Design for a Scalable Low-
Infrastructure Solution
as they maximize Currently, the scalability of hyperloop’s
core technology is a major concern for
your returns its implementation in the short-haul
from invested flight segment. It is well-known from
high-speed rail and classical maglev
building costs. projects that the infrastructure could
be as high as 95% of the total invest-
ment costs. Therefore, the possibility to
design the capsule energy-autonomous
is viewed as a very promising option. Otherwise, one
would have to electrify the whole tube, where active
rails for propulsion lead to a rapid increase in extra
components. Complex safety-critical systems along
the track favor the option of minimum infrastructure
for the propulsion. You are less interested in adding
even more costs on top of the vacuum-proof low-pres-
sure solution. It will also affect the environmental
footprint, the need for resources, the manufacturabili-
ty, and the utilization of active components. Moreover,
it is vital to make the infrastructure as affordable as
possible to ensure profitability. Thus, the self-propelled
pods are economically the best alternative, as they
maximize your returns from invested building costs. In
general, there are four alternatives for the hyperloop
propulsion system.
1) The first concept from 2013 was to make active accel-
eration zones along the track, with the rest of the
tube being passive with lower infrastructure needs. It
also made the capsule free of active propulsion com-
ponents. However, the shorter the length of the
boosting zone, the larger is the power spikes, which
will have a detrimental impact on the external bulk
power system (Musk 2013).
2) A second alternative is to design the whole tube as an
electric propulsor and tailor the needs for thrust along
the track (i.e., less needed coils for cruising). It has the
benefits of a lightweight vehicle, but the extensive
need for active components with low utilization along
the tube cannot be neglected. This option clearly
shows that there are conflicting objectives in the over-
all hyperloop design.
3) A third alternative is to combine the benefit of option
2 and apply some affordability to the solution. The
capsules are accelerated externally over a shorter
launch zone of the track, while the rest of the track is
passive, with propulsion on board to keep the cruising
speed until the capsule arrives at the destination.
4) Finally, the fully energy-autonomous option is only to
have active propulsion components onboard the
 capsule, making the tube less It affects the power voltage level of the transmission line,
infrastructure-intensive, the where costly substations of higher
track fully passive, and thus, grid with very sharp voltage levels might be required.
achieve the most affordable and System optimization is an option
scalable solution.
spikes in their highly for the power fluctuation problem,
pulsating load where the regenerative braking of
Option 1: Partially Electrified Tube capsules can be coordinated with the
with Booster Zones (Hyperloop profiles and, in the acceleration of others, but the inher-
Alpha—2013) ent complexity of such a solution is
The 2013 white paper proposed air
worst case, they are inevitable. One would also need to
bearings for suspension based on the likely to induce have some grid-side compensation
principle of aerodynamic lift. Even equipment (e.g., static VAR compensa-
though this levitation principle is voltage disturbances tors, static synchronous compensa-
mature, it is not as simple as playing tors, dynamic var devices, and so on)
air hockey inside a tube. In the hyper-
and fluctuations. to handle the pulsating loads in active
loop application, the manufacturing and reactive power. Alternatively, an
tolerances are much tighter, consider- energy storage system in between can
ing that airlift implies a low levitation height at very high be used to mitigate the highly stressful load profile, which
speeds. Moreover, the propulsion power was intended to might be one of the most cost-effective mitigation strate-
be external with 4-km sections of active stator coils along gies. It is also worth noting that the acceleration zones
the track, estimated to be long enough to increase the speed along the track require coils distributed for 4 to 5 km,
from 480 to 1,220 km/h at 1 g-force. However, rough cal- where switches can be used to energize fractions of the
culations based on the acceleration equation (Newtons’s coils close to the capsule’s position. This approach will
law of motion) show that a 5-km section would be a more improve the power factor experienced by the inverters, as
realistic figure at 1 g-force. The propulsion concept was depicted in Figure 5. The fluctuations in speed, g-forces,
using long primary linear induction motors (LP-LIMs). It and propulsive power experienced by the capsule are plot-
was proposed as a lightweight and less bulky solution for ted in Figure 6.
the capsule, which already had compressors on board. On The mean velocity for the cruising profile shown in Fig-
average, a periodic reboost of the LP-LIM would be needed ure 6 is only 798 km/h, even though the maximum speed
roughly for every 110 km (or 70 mi). As a result, less than is 1,220 km/h, yielding poor utilization of the track perfor-
4% of the tube’s overall length would need active electrifi- mance. Moreover, the repetitive spikes of 1 G also make
cation infrastructure. The kinetic energy injected during the travel a less comfortable experience.
each spike is equivalent to the energy that one ton of bat-
teries could hold (200 kWh). Table 1 presents the key met- The Hyperloop Technical Transition from Option 1
rics of the Hyperloop Alpha system (2013), and Figure 5 In the transition period between 2013 and 2018, the hyper-
depicts the booster station schematically. loop evolved toward a more traditional maglev approach,
To relieve the burden on the power grid, battery reser- where classical maglev systems are placed inside a
voirs could be installed for each
accelerator to be used for peak-
TABLE 1. The Hyperloop Alpha preliminary design (2013).
shaving. This is because the stator
segments of the LIM, including the Cruising speed 1,220 km/h (max.)—798 km/h (mean)—480 km/h (min)
converter and the power stations, Acceleration 1 g-force (i.e., 9.81 m/s2)
experience huge power spikes for
Tube pressure 0.0987% of atmospheric sea level
just a few seconds (i.e., 21 s). It
affects the power grid with very Capsule’s frontal area 1.4 m2 w/ dimensions 1.35 m × 1.10 m
sharp spikes in their highly pulsat- Capsule’s blockage 35.8% capsule-to-tube area ratio
ing load profiles and, in the worst
Capsule’s capacity 28 passengers (PAX)—2,000 kg—13.33% of total
case, they are likely to induce volt- capsule mass
age disturbances and fluctuations
Capsule’s weight 15,000 kg including PAX
(Tbaileh et al. 2021). They are also
likely to reduce a significant Energy storage on board 4,000 kg—26.67% of total capsule mass
amount of the frequency stability Propulsion system Long primary double-sided linear induction motor (LP-DB-LIM)
reserves of the grid in high-stress
Suspension system 28 air bearings—2,800 kg—18.67% of total capsule mass
situations. However, it strongly
depends on the point-of-connec- Based on https://www.tesla.com/sites/default/files/blog_images/hyperloop-alpha.pdf.
tion to the local power grid and the

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 61

 low-pressure tube. It allows a larger levitation gap than in
air bearings, which is an important safety feature at sub-
sonic speeds. In particular, maglev has 50 years of
developmental experience with actual systems work-
ing up to 600 km/h and practical lessons learned.
Currently, it is quite hard to know what is going on
inside the commercial hyperloop companies because they
are not very forthcoming in delivering the technical
details. Inspired by the SpaceX Hyperloop competition
(starting in 2017), many companies initially considered a
passive track with electrodynamic suspension (EDS) and
short-primary linear induction motors (SP-LIMs). To
improve the primitive I-beam track employed in SpaceX,
the Inductrack concept of track segmentation has been
proposed to reduce the rolling magnetic drag (proposed
by Post 2000). Hyperloop Transportation Technologies
advertises this technology (https://www.hyperlooptt
.com/technology/).
The EDS system is inherently stable and fail-safe in its
physical nature. The levitation mass on board the capsule is
based on passive Halbach-array lift skies that have no energy
requirement. However, the levitation is repulsive, which
means that the capsule has to float over the track. In addi-
tion, auxiliary wheels must be employed for takeoff and

Grid
Connection
(6 MW)

Solar Harvesting System

Step-Up
Transformer

Energy
dc dc dc Grid-Side + Storage
Converter 38 MWh
dc dc dc dc
- 37 MW-rms
56 MW-pk

dc Traction
Inverter
(70 MVA)

Solid-State
Switches

LIM LIM LIM LIM

Capsule 480 km/h 1,220 km/h Capsule

1,220 km/h Capsule 480 km/h

LIM LIM LIM LIM

Solid-State
Switches

dc Traction
Inverter
(70 MVA)

Figure 5. A schematic sketch of the acceleration booster concept (Hyperloop Alpha system). (Source: Musk 2013.) The integration with the
power system and the tube’s harvesting of solar electricity is also depicted.

62 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

landing (i.e., no suspension available at low speed). Moreover,
the large levitation height of the Inductrack EDS makes it
incompatible with linear propulsion, which means that the
linear motor has to be separate and act over another track
surface. If one decides to make the propulsion external
instead of on board the capsule, the linear synchronous
motor (LSM) is considered the most energy-efficient solution.
However, the LSM favors electromagnetic suspension (EMS),
as these technologies can be integrated with each other. This
suspension concept relies on attractive levitation rather than
the repulsive nature of the EDS. With EMS, the hyperloop
capsule could be hanging underneath the track inside the
tube. It is postulated by the companies that a hanging cap-
sule could improve cornering when it turns inside the tube.
Still, the radius of curvature would need to be much higher
for the hyperloop than classical rail to restrict the centrifu-
gal g-force experienced by passengers (e.g., a 0.2 g-force
comfort level implies a curvature radius of 39.3 km at
1000 km/h cruising speed). In 2018, Virgin Hyperloop (VH)
switched its strategy from the Induc-
track to the EMS to integrate its sus-
pension with LSM propulsion. 1,200
1,000
Speed (km/h)
Option 3: Electromagnetic Launch System
With Self-Propelled Cruising
As illustrated earlier, a long track-length is needed for
acceleration if the propulsion power level is not sufficient-
ly high enough. This is where the concept of the tube as a
propulsor (i.e., option 2) finds its benefit with excellent
acceleration performance. For this reason, the Spanish
hyperloop company Zeleros has proposed a similar sys-
tem as an electromagnetic launch system during accelera-
tion (LSM as a track propulsor), even though it is pursuing
a self-propelled capsule in the cruising zone (https://zeleros
.com/hyperloop-technology/). In this way, its capsule is
saving a massive amount of energy needed for accelera-
tion [e.g., potentially reducing the energy reservoir’s mass
with up to 50% (Nøland 2021)], which makes its vehicle
lighter than a fully energy-autonomous solution. It also
reduces the tube length needed for acceleration, but at the
expense of higher complexity compared to a fully self-
propelled solution. Contrary to all of the other commercial

Option 2: Large-Scale Tube 800

Electrification With a
600
Lightweight Capsule
The concept of the rail as a propul- 400
sor was popularized by the German 200
Transrapid maglev system. Howev- 0
er, it has been concluded that the 1.2
reason the system failed was that 1
the guideways were too expensive, 0.8
even though the ride itself was per-
G-Force

0.6
fect. VH and Hardt are now pursu- 0.4
ing this option for the hyperloop,
0.2
even though it is well-known that it
0
has massive infrastructure needs.
−0.2
A basic sketch of the system is
50
Acceleration Power (MW)

presented in Figure 7.
The track propulsor can indeed 40
be tailored to the propulsion needs
30
along the track, and therefore,
20
needs less powerful components in
the cruise zone of the tube. The 10
LSM can achieve very good efficien- 0
cy and power factor, given that the
0
5

118
123

236
241

354

system only energizes the portion

of the track where the capsule is
situated every time instant. Due to
no slip between the primary and
the secondary, it can reach the
highest cruising speeds for a given
power supply, contrary to LIMs.
However, the need for synchronism
implies higher complexity in how
the system is operated.
Distance (km)
Figure 6. The speed profile, G-force, and acceleration power of the Hyperloop Alpha passenger
capsule, with a total weight of 15 tons. Maximum and minimum speeds are 1,220 km/h and 480 km,
respectively. The acceleration force is 1 G over the boosting zone. In between the acceleration
spots, aerodynamic forces are acting to deaccelerate the capsule, proportional to the square of
the instantaneous speed. The mean electrical power absorbed during acceleration is 34.74 MW.
The capsule uses approximately 21 s inside the boosting zone and 9 min and 9 s in between.
Note that 202.2 kWh kinetic energy is injected under each spike, while discharged the same
amount of energy in between due to natural air resistance and friction. The boosting interval
repeats periodically through the whole cruising zone of the track.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 63

 companies, Zeleros considers aerodynamic propulsion in
the cruising zone, powered by an electric rotary motor rath-
er than a linear one. A rotating propulsor will have a very
high efficiency because it has a much smaller air gap than a
linear induction motor. However, even though the electrical
losses are significantly reduced and will be easier to handle,
losses are now manifesting in aerodynamic propulsion,
which will require extra power instead. Another problem is
that the aerodynamic thrust highly depends on the tube’s
air density, causing competing interests in the design. By
selecting a tube operated at the Concorde subsonic aircraft’s
commercial pressure levels (1/10 of sea level), the air

Longitudinal View
Zone Pitch

N N N
Guideway (Primary)

Pole Pitch
+
Capsule (Secondary)
-

Cross-Sectional View

Guideway (Primary)

Linear Synchronous Linear Synchronous

Motor Stator Motor Stator

Hybrid Hybrid
EMS EMS

Capsule (Secondary)

Lateral Lateral
Guidance Guidance
Suspension Suspension

Figure 7. The large-scale tube electrification solution with a lightweight capsule. The concept is depicted in both a longitudinal and a cross-sec-
tional view, respectively. The propulsion system is distributed along the track as a long primary LP-LSM. The levitation is an electromagnetic sus-
pension (EMS) system with hybrid excitation. The concept refers to the solution proposed by Hardt Global Mobility (https://hardt.global/
technology-development/).

64 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

resistance will be 100× higher than in a tube operated at
1/1000 of the atmospheric pressure but only 40% of the drag
experienced by today’s commercial aircraft.
Option 4: Energy-Autonomous Capsule
With a Low-Infrastructure Tube
Even though Zeleros proposes a partially self-propelled
and scalable solution, Transpod is currently the only com-
pany going for a fully energy-autonomous transporta-
tion system (https://www.transpod.com/technology
-demonstrator/). It implies very low construction costs,
which means that if the technology is successful, it can
further improve the profitability and affordability of the
hyperloop. Basically, this type of system configuration is
also the typical system explored in the SpaceX Hyperloop
competition for students. An example of such a conceptu-
al system is shown in Figure 8. It is worth noting that
SpaceX uses an I-beam track and that Transpod does not
have passive lift skies. In Transpod’s technology, the linear
motor is also utilized for levitation, guidance, and braking,
proposing a very challenging all-in-one solution. More-
over, to cope with the limitations in onboard energy stor-
age, Transpod proposes a contactless high-speed power
transmission system. The proposed wireless technology

View From Above

dc

N1

-
+
Guideway (Secondary) -
+
N2

dc
Inductrack Array

Cross-Sectional View

Capsule (Primary)

Lift Ski Double Sided LIM Lift Ski

Coil Back Back Coil
Iron Iron hc
h Wood Wood
Guidance Guidance
Aluminum Track

Guideway (Secondary)
w

Figure 8. The energy-autonomous capsule solution implying low-infrastructure needs and less electrification along the track. The concept is
depicted in both view from above and cross-sectionally, respectively. The propulsion system is a short-primary double-sided linear induction motor
(SP-DS-LIM). It is primarily installed on board the capsule, while the track is passive with no active components. The levitation is configured as a
passive EDS system with auxiliary lift-off wheels and no feedback control needed. The concept refers to a similar solution as proposed by Trans-
pod (https://www.transpod.com/technology-demonstrator/), where the passive lift skis were omitted.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 65

 TABLE 2. An assessment of the two main hyperloop concepts.
Energy-Autonomous Capsule With a Scalable
Concepts Low-Infrastructure Tube
Tube Infrastructure Capsule is self-propelled—it is the best option
in terms of affordability, using of few active
components
Capsule weight High—energy storage for onboard propulsion
with or without contactless power transmission
Range limitation Depends on the onboard energy reservoir or
the feasibility of a wireless power supply while
moving
Acceleration Power and energy allocated for acceleration af-
limitation fects the range even if regenerative braking
under deacceleration is achieved
takes into account the low pressure inside the tube to
potentially maximize the power transmission efficiency.
Currently, one of the most difficult issues to solve for
the self-propelled system is the management of the heat
losses on board. The linear induction motor’s efficiency
could be as low as 70% during cruising, which implies that
a significant portion of the inverter rating will feed losses
(Nøland 2021). The white paper from 2013 proposed an
onboard water tank to absorb all of the waste heat and
replace the water when it arrives at the station. Unfortu-
nately, this type of solution will add to the total mass of
the capsule. It is also difficult to throw heat waste out of
the capsule and into the tube because of the lack of air
density. As a result of the lack of air, the convective heat
transfer is significantly reduced, and the radiation part is
dominating. A summary of the two key concepts of exter-
nal and self-propulsion is provided in Table 2.
Conclusion
This article introduced some of the most recent technology
evolutions of the HTS, intended to make it feasible, scalable,
and affordable for implementation. In particular, it can be
perceived that the external propulsion enables a lightweight
capsule and might have a faster technical development track
to realization and commercialization. However, a self-pro-
pelled capsule configured like an airplane requires less track
infrastructure and utilizes its active components during the
whole journey. While its low construction costs would signif-
icantly improve the system’s profitability and reduce the
maintenance of the operated infrastructure, the capsule
tends to get heavy when considering the onboard energy
storage and thermal management system. Therefore, com-
panies are now considering a hybrid solution, combining
external launching and self-propelled cruising, which bal-
ances the benefits and drawbacks of both variants.
In addition to the technical challenges and opportunities,
there are also societal changes and policy decisions that
might play a role in speeding up hyperloop implementation.
There is currently a push to ban short-haul domestic flights
66 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
Large-Scale Tube Electrification With a
Lightweight and Externally Driven Capsule
Active rails with significant infrastructure cost,
where active components are only utilized during a
tiny fraction of the ride
Low—lightweight capsule with only onboard auxiliary
energy (e.g., for controlling suspension)
Unlimited, as the propulsion is external but high
investments and the resource intensity favor short
range
Acceleration power can be high and the distance
needed for acceleration can be low but creates power
spikes for the grid
in Europe to accelerate the decarbonization of transport. An
alternative to rail is introducing hyperloop, which has the
potential to significantly decrease the energy use per RPK
when compared to aviation, and at the same time, move
with similar or higher travel speed. Still, no full-scale HTS
has yet been demonstrated at subsonic or near-sonic
speeds. However, this article tries to give more insight into
where the development is going and make predictions on
the future realization of this new mode of transportation.
For Further Reading
Communication from the Commission to the European Par-
liament, the Council, the European Economic and Social Com-
mittee and the Committee of the Regions, “Sustainable and
Smart Mobility Strategy—putting European transport on track
for the future,” European Commission, Brussels, Belgium, Dec.
9, 2020. [Online]. Available: https://ec.europa.eu/transport/
sites/transport/files/legislation/com20200789.pdf
J. K. Nøland, “Prospects and challenges of the hyperloop
transportation system: A systematic technology review,” IEEE
Access, vol. 9, pp. 28,439–28,458, Feb. 2021. doi: 10.1109/
ACCESS.2021.3057788.
E. C. Goddard, “Vacuum tube transportation system”, U.S.
Patent 2 511 979, June 1950.
E. Musk, “Hyperloop alpha,” Hawthorne, CA, White Paper,
2013. [Online]. Available: https://www.tesla.com/sites/default/
files/blog_images/hyperloop-alpha.pdf
D. Tudor and M. Paolone, “Optimal design of the propulsion
system of a hyperloop capsule,” IEEE Trans. Transport. Electrif-
ic., vol. 5, no. 4, pp. 1406–1418, Dec. 2019. doi: 10.1109/
TTE.2019.2952075.
A. Tbaileh et al., “Modeling and impact of hyperloop tech-
nology on the electricity grid,” IEEE Trans. Power Syst., vol. 99,
pp. 3938–3947, Mar. 2021. doi: 10.1109/TPWRS.2021.3056298.
R. F. Post, “Maglev: A new approach,” Sci. Amer., vol. 282, no.
1, pp. 82–87, Jan. 2000.
Biography
Jonas Kristiansen Nøland (jonas.k.noland@ntnu.no) is
with the Department of Electric Power Engineering at the
Norwegian University of Science and Technology, Trond-
heim, 7491, Norway.

 By Federico Lluesma, Antonio Arguedas, Sergio Hoyas,
Alberto Sánchez, and Juan Vicén

ODAY, EVACUATED-TUBE

T
transport (ETT ) sys-
tems, also referred to
as hyperloops, are becom-
ing ever more famous.
They consist of a ground-based net-
work enclosed within a tube in
which the atmosphere can be con-
trolled. If the air inside the tube is
evacuated, a low-pressure environ-
ment is created and aerodynamic
resistance can be reduced, enabling
higher speeds and efficient energy con-
sumption. A hyperloop adds one element

©S
HU
T
to an ETT: levitation. As a result, no ground

TER
ST
OC
resistance exists, overcoming one of the

K.
CO
M
/K
IT8

major limitations of traditional trains.

.N
ET

Evacuated-Tube,
High-Speed,
Autonomous
Maglev (Hyperloop)
Transport System for
Long-Distance Travel
An overview.

Digital Object Identifier 10.1109/MELE.2021.3115543

Date of current version: 1 December 2021

2325-5897/21©2021IEEE IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 67

 Origins of ETT
The history of ETT begins more than
200 years ago when, in 1799, English
inventor George Medhurst proposed
moving passenger carriages through
a tunnel by using variations in pres-
sure levels. Systems were built in
France, England, and Ireland during
the 19th century and called “atmo-
spheric railways.” Even then, people
were concerned about ETT, as dem-
onstrated by Figure 1.
A century later, in 1904, Robert
Goddard, at Worcester Polytechnic
Institute, Massachusetts, combined
the concept of airless tunnels to
reduce resistance with the idea of
magnetic levitation (maglev) sys-
tems to reduce ground friction losses. This could
enable very high speeds with relatively little power for
propulsion. Russian inventor Boris Weinberg built his
first model, in 1909, at Tomsk University. During the
following years, the evolution of both technologies
(low-pressure tubes and maglev systems) advanced in
parallel. The development of the first superconduct-
ing levitation technologies and linear motors led to
the creation of maglev trains. Concurrently, advances
in space and pipeline construction resulted in the
development of high-volume vacuum systems, whose
preeminent example is the Large Hadron Collider,
Figure 1. A satiric cartoon from an English newspaper in 1828, showcasing transport inventions. [Source: Reproduced under a Creative Com-
mons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license. © The Trustees of the British Museum.]
68 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
Although hyperloops with a total of 104 km of piping
under vacuum.
were proposed more In the late 1970s, Prof. Marcel Jufer,
from Switzerland, combined pressure
than a century ago, and magnetism, proposing Swissmet-
they have become ro to operate at pressures at which
the Concorde SST was certified to fly,
famous only during but no operational line was imple-
mented. In August 2013, American
the past six years, businessman Elon Musk published an
due to a need to online whitepaper, “Hyperloop
Alpha,” with a concept for air bear-
develop their ings for levitation and linear motors
for propulsion at every station. In
technologies. 2015, Musk’s SpaceX organized a uni-
versity competition for students
around the world to share ideas to
boost hyperloop evolution. The idea
was to develop an annual event to motivate the develop-
ment of innovative ideas and recruit the best students for
the company.
From 2015, hyperloop companies arose globally to
develop such a system, and cooperation agreements have
been announced to enable international standards ensur-
ing interoperability. During this time, a team at Universitat
Politècnica de València (UPV), Spain, Hyperloop UPV, was
created. It aimed to compete against top technological
universities around the world and develop a hyperloop.
After winning two prizes for its first prototype (best overall
design and best propulsion system), it decided to go one
step further and create Zeleros, an independent company.
Its hyperloop concept appears in Figure 2.
Key Benefits of Hyperloop
Transportation Systems
Why Now?
Although hyperloops were proposed more than a century
ago, they have become famous only during the past six
years, due to a need to develop their technologies, includ-
ing the following:
xx Vacuum: Hyperloops require large vacuum chambers.
Today, the Large Hadron Collider, which began opera-
tion in 2009, accounts for 54 km of ultrahigh vacuum
(around 10−10 mbar).
xx Maglev: The first maglev train operated at the Bir-
mingham, U.K., airport from 1984 to 1995. The tech-
nology is sufficiently mature to facilitate high-speed
maglevs in China and Japan.
xx Control: High-frequency automation has been put to
the test in several applications, such as the reusable
Space X Falcon rocket, which is able to land vertically
after completing a mission.
xx Composite and intelligent materials: These enable build-
ing light and extremely resilient fuselages.
xx Power electronics and drives: These devices are well
developed, but there is still a wide margin for
improvement in the railway sector. Progress is occur-
ring in the automotive industry and aviation, where
there are clear advances through projects such as the
E-Fan X from Airbus.
Figure 2. Zeleros’s hyperloop.
Apart from technology, the necessity of a new means
of transport is becoming more and more evident. Trends
show ever-lengthening travel distances, particularly where
high-speed trains and planes share or overlap markets. Air
space congestion is especially worrying. By 2050, Airbus
expects the number of flights to increase 100%. Eurocon-
trol, the organization that manages the air space in
Europe, says in one of its reports, “For the future, with the
saturation of airspace resources in the long-term, the con-
gestion problem cannot be overcome unless the airspace
structure is reorganized.”
In addition, as 21st century systems, hyperloops would
be the greenest means of transportation: they are fully
electric and hence have zero direct emissions. Thus, they
are proposed as a solution to transportation saturation, as
they would form a complimentary transport layer that can
contribute to reducing pollution. Demand for air travel will
continue growing. Deploying hyperloops could alleviate
congestion by adding ground systems with performance
similar to that of airplanes.
Description of a Hyperloop
As described, hyperloops have several advantages. On
the one hand, they are based on levitation. Thus, high-
speed railway problems, such as wheel hunting and
expensive rail maintenance, can be resolved, and effi-
ciency can be improved. Apart from that, the partial
vacuum inside the tube reduces the aerodynamic drag
on a train. Also, the elimination of friction enables
these systems to be as fast as planes, with the sound
barrier as the theoretical limit (around 1,250 km/h).
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 69
 That is twice the fastest speed obtained by high-speed
rail and maglevs and similar to that of most fighter jets,
with considerably less energy consumption (see Figure
3). Tubes also offer protection from adverse weather
conditions. Crosswinds, heavy rain, and snow would
have less impact on hyperloops. Numerous approaches
have been taken to develop hyperloops, and several
companies have emerged. Zeleros is one of them,
whose approach is described in the “Description of
Zeleros’s Hyperloop” section.
One of the drawbacks of hyperloops could be the
amount of energy consumed by the pumps to maintain
low pressure inside the tube. On the other hand,
assuming that electric power of 40 W/m 2 can be
extracted from solar panels and that the tube can be
encased in panels, a 4-m diameter tube would produce
160 kW/km, which is, for an initial computation, more
than enough to energize the pumps in a realistic sce-
nario. Finally, the vehicles will be autonomous. This
enables operation from a control center, eliminating
the human factor from safety concerns and helping to
reduce operating costs, including long training periods,
simulators, and manuals.
800
Energy Per Seat (MJ)
Description of Zeleros’s Hyperloop
The Zeleros concept (Figure 2) can be divided into two
systems. The levitation is based on hybrid electromagnet-
ic suspension. Permanent magnets surrounded by elec-
tromagnets are placed on the vehicle. They lift the weight
of the vehicle, while the electromagnets control the gap
between the vehicle and tube wall. This way, less power is
required to energize the coils, as most of the work is done
by the permanent magnets. On the other hand, the main
propulsion is based on a simplified aeronautical engine
from which the combustion chamber has been removed.
A compressor driven by an electric motor captures the air
in front of the pod, compresses it, and exhausts it through
a nozzle to generate the required thrust.
One of the main advantages is that, as all the technolo-
gy is integrated into the vehicle, the track complexity is
significantly reduced compared to other approaches that
base their propulsion on linear motors. Since no coils and
permanent magnets are required on the track, the infra-
structure costs considerably less. Another benefit is that
excess energy in the flux can be recovered using a turbine,
making the whole system more efficient.
Regarding speed, there is a debate not only from the
technical perspective but also
from the customer and user
point of view. At a minimum,
any transportation system
added to the current portfolio

600
needs to provide better service
than existing ones. The fastest
400
means of travel for conven-
tional passengers is the air-
200 plane. In addition, it is well
known that commercial air-
0 planes are the safest mode of
Paris–Frankfurt Boston–Baltimore New Delhi–Ahmedabad transportation. So, hyperloops
600 km 700 km 900 km should at least match air-
planes in these two features.
Figure 3. The energy consumption per seat for different transport modes (one way). (Sources: Siemens The problem to be solved is
and Flight Aware.) not what a hyperloop’s top
speed would be but what its
average speed would be for a
TABLE 1. The duration of flights from Barcelona, Spain, to European given corridor.
cities. (Source: Flight Aware.) An example of airplane
average speeds can be found in
Distance (km) Distance (km) True Speed Real Travel Time Table 1; it demonstrates that
Destination (Straight Line) (Actual Flown) (km/h) (Including Taxi)
despite having top speeds of
Toulouse, France 267 332 398 53 min 850–950 km/h, planes’ average
speed (their so-called true
Paris, France 828 900 452 1 h, 48 min
speed, which includes taxi
Brussels, Belgium 1,085 1,259 567 2 h, 7 min travel time), is much lower, as
Amsterdam, 1,243 1,378 498 2 h, 26 min graphed in Figure 4. Consider-
The Netherlands ing this, there is a case to be
Copenhagen, Denmark 1,771 1,996 591 2 h, 57 min made for a ground system that
can sustain a cruising speed of
Stockholm, Sweden 2,317 2,530 680 3 h, 32 min
at least 650 km/h for distances

70 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

at which it remains competitive with planes (fewer than
3 h). What if the cities in the table were linked through a
single hyperloop main line? The issue is whether it is pref-
erable to take a 3-h, 30-min flight from Spain to Sweden or
a 4-h hyperloop ride, assuming a sustained 650 km/h. This
example does not factor in time spent in stations and air-
ports, which has become a drawback to air travel, and it
does not consider the fact that hyperloop stations could be
placed in city centers. Safety will be discussed in more
detail in the following, but a quick note is that hyperloops
require no takeoffs and landings.
Electric Propulsion for Hyperloops
As explained, hyperloops are based on vehicles on closed
tracks in low-pressure tubes that minimize energy con-
sumption while operating safely and reliably. The purpose
of the propulsion system is to achieve this and overcome
the natural drag that the fluid inside the track exerts on
the vehicles and other resistance, such as magnetic drag.
The vehicles will transmit kinetic energy to the surround-
ing fluid, dragging a wake behind them. If the pod veloci-
ty is high enough, shock waves will occur, similar to those
that take place in a high-speed train tunnel. However,
shock waves in hyperloops cannot be released into the
atmosphere, as they are for trains. They will bounce back
and forth, hitting the vehicles and potentially jeopardiz-
ing stability.
Several hyperloop concepts use linear motors all along a
route in a complete vacuum tube to avoid this effect. That
level of vacuum may require a more complex and delicate
two-stage pump. The high vacuum is required to create
conditions where there is no aerodynamic drag. Neverthe-
less, the adverse effects of atmospheric pressure inside the
tube can be overcome with a potentially better solution, an
electric turbofan-like propulsion system. Turbofan engines,
which are the most common for short-, medium-, and
800
700
600
452
500 398
Speed (km/h)
400
300
200
100
0
0 250 500 750 1,000
Distance From Barcelona (km)
Figure 4. The true speed of flights from Barcelona, Spain, operated by Vueling Airlines on 17 November 2019, using an Airbus 320 aircraft.
(Source: Flight Aware.)
long-haul aircraft, are following the road map of other tradi-
tional propulsion systems, such as cars. Car engines, where
mechanical power was traditionally generated by burning a
hydrocarbon fuel, are being replaced by electric power units
fed from batteries. Turbofan engines with electromechani-
cal power units have the advantage of producing no emis-
sions. This enables their use in confined facilities without
incurring recirculating gas problems.
The use of an electric turbofan-like system is mandato-
ry when the pressure inside the tube is at nonspace vacu-
um levels, which is basically due to vehicle aerodynamic
drag. This system has two main effects: first, there is a
way to cope with the piston effect. Air is transferred after
a compression–decompression cycle from the front to the
tail of the vehicle. Second, a propulsion system based on
compressed air is obtained, compensating for the higher
drag. A value of 10 kPa is enough to reduce the aerody-
namic drag by more than one order of magnitude and
enables the compressor to properly operate.
Electric turbofan systems have a disadvantage in that
they are fit to operate only at their design point in an
enclosed environment. However, at the design point,
hyperloops are extremely competitive from an energy
efficiency point of view, compared to aircraft and piston
engine vehicles, which operate outside their design con-
ditions for significant amounts of time. For short-haul
routes, meaning fewer than 1,500 km, an aircraft is at its
cruising speed for only about 70% of the distance (Fig-
ure 5). The efficiency of a system with such a characteris-
tic is quite poor. Considering energy, a significant portion
is wasted, especially during taxiing, takeoffs, and climb-
ing (Figure 6). Hyperloops have the advantage that for
routes in this range, they can reach their travel speed and
maximum efficiency without a phase equivalent to
climbing, which increases their cruising phase to approxi-
mately 95% of a trip.
680
591
567
498
1,250 1,500 1,750 2,000 2,250 2,500 2,750
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 71
 Depending on the size of a hyperloop’s cargo and pas-
Speed (km/h)
1,200 senger cabin, the mass air flow that the electric turbofan
900 takes in may have to be compressed beyond the desired
600 pressure ratio for propulsion efficiency. For such cases,
300 regeneration through an air turbine is required to achieve
0 high-efficiency values. Hyperloop vehicles must have very
Route
high overall efficiency, otherwise, energy that is not used
Figure 5. A daily flight from Delhi to Ahmedabad, India (900 km), for propulsion will be introduced into the track environ-
spends less than 70% of the time cruising. ment, altering the optimum operating conditions. From a
safety and reliability point of view,
higher pressures facilitate these
Peak Power Required (MW)

40 systems with confidence at 10 kPa.

35 The Concorde proved that pressur-
30 ized cabins and turbofan engines
25
could operate at pressures lower
20
15
than 10 kPa (Figure 7). The technol-
10 ogy for operating at those pres-
5 sures has a high level of maturity,
0 which guarantees its use for pas-
Taxi Out Takeoff Climb Cruise Descent Approach Taxi In senger and cargo transportation.
and Land
Below these pressures, the ineffi-
Figure 6. The power required for a generic, 150-passenger, short-haul aircraft designed for a max- ciencies of turbomachinery are
imum range of 3,000 nautical miles and a 1,000-nautical-mile (1,800-km) mission. well studied, and a method to
avoid the piston effect is required.
From a turbomachinery and
Sea Level pressurized cabin point of view, the
1,000 Concorde’s roof could be established
for passenger and cargo transporta-
tion as a certified limit. In terms of
Pressure (mbar)

100 Armstrong’s Limit safety, a boundary could be estab-

lished at the Armstrong limit of
6.26 kPa. Operating at vacuum levels
10 lower than the Armstrong limit is
potentially beneficial for energy
consumption, given the almost com-
1 plete mitigation of adverse aerody-
Aircraft Concorde Zeleros SR-71 Balloon Hyperloop
namic effects. However, cabins
Commercial Supersonic Hyperloop Military Red Bull Low-Pressure
Commercial High-Pressure Aircraft Stratos Concepts certified to operate at those pres-
Aircraft Concepts sures have specific and restrictive
safety protocols that may be unfeasi-
Figure 7. The operating pressures for different vehicles. (Sources: SR-71 Online and Red Bull.) ble for massive passenger transpor-
tation systems. Currently, vehicles
operating below the Armstrong limit
6 cannot ensure survivability in the
event of depressurization, making
5
Total Travel Time (h)

this approach riskier. Typically, the

4 threat is mitigated using space suits,
3 such as those for high-altitude
2
flights, that provide military pilots
and astronauts a unipersonal breath-
1 ing environment.
0 All hyperloop systems will benefit
Paris–Frankfurt Boston–Baltimore New Delhi–Ahmedabad
from progress in electrical technolo-
600 km 700 km 900 km
gy to offer a zero-direct-emission
Figure 8. Point-to-point times for different transport modes. (Sources: SNFC, Deutsche Bahn, system. They will be designed from
Indian Railways, Amtrak, and Flight Aware.) their foundations to provide green

72 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

transportation. They will not only be fossil fuel free but also
quiet, since most or all noise produced will be kept inside the
tube. Solar panels and windmills will be the preferred source
of energy whenever possible.
Main Operational Characteristics
of the Proposed Hyperloop System
It has been proved through simulation that hyperloop vehicles
can be built to accommodate up to 200 passengers, so they are
being designed to carry between 50 and 200 people. The larg-
est and most common regional airplanes, the Airbus A320 and
Boeing 737, can carry up to 185 passengers. Thus, the system
is aligned with true market needs in this respect. Also, for the
proposed range and design, the more passengers a vehicle
carries, the more efficient it becomes in terms of consumption
per seat. Nonetheless, vehicle capacity can be adjusted to spe-
cific needs, accommodating 50, 75, 100, or 130 riders. Most like-
ly, the first to be deployed will carry 50 passengers, enabling
power requirements to be more easily be met and providing
enough capacity for most of the potential corridors. Consider-
ing a single tube per direction and 16 h of daily operation, and
a tentative minimum headway at a cruise speed of around
5 km/h, 38,400 passengers per day per direction could be
accommodated, with a flow of 2,400 passengers during peak
hours. For a point-to-point network targeting distances longer
than 500 km, these figures are sufficient in most cases, and
there is still a reasonable margin for improvement.
Furthermore, one key benefit of the proposed system is
that it could complement air traffic. It has the potential to
add highly required capacity in already congested corri-
dors. Also, a dynamic simulator was developed to estimate
the cost in time and energy that Zeleros’s concept
requires. The results are compared with planes and trains
for different routes on three continents (Europe, North
America, and Asia). Each is representative of distances
ranging from 600 to 900 km. In Figure 3, it can be seen that
the energy consumption is far closer to that of trains,
being more than two times less than that of planes. In Fig-
ure 8, the travel time matches planes at a cruise speed of
650 km/h and is far better than any rail service covering
the same distance.
Current State of Development and Next Steps
Several private companies are involved in hyperloops:
Zeleros; Hardt, in The Netherlands; Nevomo, in Poland;
Transpod, in Canda; and Virgin Hyperloop and Hyperloop
Transportation Technologies, in the United States. Thanks
to rapid growth and maglev initiatives in Asia, almost all
of the hyperloop system has been tested on a laboratory
scale. The larger development has been done by Virgin
Hyperloop. In its 500-m tube, a capsule reached 387 km/h
in 2017, using magnetic propulsion and passive maglev in
a low-pressure environment. Later, in 2020, a similar cap-
sule transported two people up to 173 km/h.
Although almost all the technologies have been tested,
major challenges remain in how to integrate them in an
efficient and commercially operable vehicle. The behavior
of the magnetic system at 1,000 km/h (for levitation and
propulsion) is an uncertainty, as the fastest maglev speed
to date is 603 km/h, reached on the Japanese Shinkansen
in 2015. Other issues such as cabin conditioning, emer-
gency evacuation protocols, and vehicle flow at stations
are unresolved. Apart from technology development,
there has been a huge step forward in terms of certifica-
tion. In the European Commission, a standardization
committee has been created, and the requirements for
this means of transportation are being developed.
Acknowledgment
We wish to acknowledge all the people involved in this
project, including those advisors, investors, part-
ners, and team members who believe in a better future
for everyone.
For Further Reading
H. Clayton, The Atmospheric Railways. Gloucester: Alan Sutton
Publishing, 1985.
“Hyperloop pioneers welcome a new era in high-speed
transportation with industry-first international alliance,” EIT
Climate-KIC, 2018. https://www.climate-kic.org/news/zeleros-
hyperloop-consortia/ (accessed May 31, 2019).
G. A. Landis, “Human exposure to vacuum,” Aerospace Web,
July 7, 2009. http://www.geoffreylandis.com/vacuum.html
(accessed Feb. 5, 2016).
“Birmingham airport’s old Maglev carriage to be sold,”
BBC. https://www.bbc.com/news/uk-england-birming-
ham-11780663 (accessed Nov. 26, 2019).
“Hyperloop pod competition,” SpaceX. https://www
.spacex.com/hyperloop (accessed Nov. 27, 2019).
“At 574.8 km/h, the TGV,” Le Parisien. http://www.leparisien
.fr/archives/a-574-8-km-h-le-tgv-04-04-2007-2007914213.php
(accessed Nov. 27, 2019).
A. A. Pilmanis and W. J. Sears, “Physiological hazards of
flight at high altitude,” Lancet, vol. 362, no. 1, pp. 16–17, Dec.
2003. doi: 10.1016/S0140-6736(03)15059-3.
“The Japanese Maglev: World’s fastest bullet train,” Japan
Rail Pass, Nov. 2019. https://www.jrailpass.com/blog/maglev
-bullet-train
National Academies of Sciences, Engineering, and Medi-
cine, Commercial Aircraft Propulsion and Energy Systems Research:
Reducing Global Carbon Emissions. Washington, D.C.: The
National Academies Press, 2016.
Biographies
Federico Lluesma (fellrod@doctor.upv.es) is with Universi-
tat Politècnica de València, València, 46022, Spain, and
Zeleros, València, 46024, Spain.
Antonio Arguedas (aarguedas.paths@outlook.com) is
with Zeleros, València, 46024, Spain.
Sergio Hoyas (serhocal@upvnet.upv.es) is with Univer-
sitat Politècnica de València, València, 46022, Spain.
Alberto Sánchez (Alberto.sanchezgimenez@gmail.com)
is with Zeleros, València, 46024, Spain.
Juan Vicén (jvicen@zeleros.com) is with Zeleros, Valèn-
cia, 46024 Spain.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 73
 DAT E S A H E A D
The safety and well-being of all conference participants
are our priority. Our conference organizers continue to
monitor official travel advisories related to the corona-
virus (COVID-19) pandemic. Please check the event
websites for the latest information.
2021
19– 22 D E C E MB E R
STPEC 2021: IEEE Second International Conference
on Smart Technologies for Power, Energy, and Control,
Bilaspur, India, http://www.stpec2021.in/
2022
2– 5 J AN U ARY
PESGRE 2022: IEEE International Conference on Power Elec-
tronics, Smart Grid, and Renewable Energy, Thiruvanantha-
puram, India, http://pesgre2022.org/
9– 12 J AN UARY
JTCM 2022: IEEE PES Joint Technical Committee Meeting,
Orange County, Garden Grove, California, United States,
contact Solveig Ward, sward@quanta-technology.com,
https://pestechnical.org/
21– 24 F E BR U ARY
ISGT 2022: IEEE PES Innovative Smart Grid Technologies,
Washington, D.C., United States, contact Kathy Heilman,
kathy.heilman@ieee.org, https://ieee-isgt.org/
20– 24 MAR C H
APEC 2022: IEEE Applied Power Electronics Conference and
Exposition, Houston, Texas, United States, http://www.apec
-conf.org/
25– 28 AP RIL
T&D 2022: IEEE PES Transmission and Distribution Confer-
ence and Exposition, New Orleans, Louisiana, United States,
contact Carl Segneri, carlsegner@sbcglobal.net, http://www
.ieeet-d.org/
28– 30 AP RIL
IFEC 2022: IEEE International Future Energy Challenge,
Knoxville, Tennessee, United States, contact Dr. Hua Bai,
hbai2@utk.edu, http://energychallenge.weebly.com/
ifec-2022.html
Digital Object Identifier 10.1109/MELE.2021.3115544
Date of current version: 1 December 2021
74 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
1 –5 MAY
IAS/PCA 2022: IEEE-IAS/PCA Cement Industry Conference,
Las Vegas, Nevada, United States, http://cementconference.org/
2 –5 MAY
I&CPS 2022: IEEE/IAS 58th Industrial and Commercial Power
Systems Technical Conference, Las Vegas, Nevada, United
States, http://attend.ieee.org/icps-2022/welcome/
1 5 –1 9 MAY
IPEC-Himeji 2022-ECCE Asia: International Power Electron-
ics Conference, Himeji, Japan, http://www.ipec2022.org/
1 5 –1 7 JUNE
ITEC 2022: IEEE Transportation Electrification Conference
and Expo, Anaheim, California, United States, contact
Rebecca Krishnamurthy, rebecca.k@rna-associates.com,
https://itec-conf.com
1 7 –2 1 JULY
GM 2022: IEEE PES General Meeting, Denver, Colorado, United
States, contact Roseanne Jones, roseanne.jones@ieee.org,
https://pes-gm.org/
5 –9 S E P TE MBE R
EPE’22 ECCE Europe: 24th European Conference on Power
Electronics and Applications, Hanover, Germany, http://
www.epe2022.com/
1 2 –1 5 S E P TE MBE R
PowerCon 2022: IEEE International Conference on Power
System Technology, Kuala Lumpur, Malaysia, contact Zuhai-
na Zakaria, zuhaina@gmail.com, https://attend.ieee
.org/powercon-2022/
For more information on upcoming conferences, please
visit the following websites:
IEEE Power & Energy Society: https://www.ieee-pes.org/
meetings-and-conferences/conference-calendar
IEEE Transportation Electrification Community: https://tec
.ieee.org/conferences-workshops
IEEE Industry Applications Society: https://ias.ieee.org/
events-conferences/conference-schedule.html
IEEE Power Electronics Society: https://www.ieee-pels.org/
conferences
 Save the Date

JOIN US for the 2022 IEEE Power &

Energy Society General Meeting

Denver, Colorado | July 17–21, 2022

The IEEE PES General Meeting is the premier global technical

conference for the electric power and energy community,
attracting over 3,500 attendees from more than 70 countries
from all over the world.

• Connect and network with industry associates & peers

from across the globe.
• Enjoy technical tours of local industry corporations &
facilities.
• Be engaged during one of our leading technical
presentations or panel discussions led by prominent
industry professionals.
• Bonus! Conference attendance accounts for professional
development hours (PDHs).

Digital Object Identifier 10.1109/MELE.2021.3126974 pes-gm.org

 NEWSFEED
ITEC+: The IEEE Transportation
Electrification Conference and
Expo and AIAA/IEEE Electric Aircraft
Technologies Symposium
OR 10 YEARS, THE IEEE
F Transportation Electrifica-
tion Conference and Expo
(ITEC) has served as IEEE’s premier
conference on transportation electri-
fication, fostering connections
between industry and academia. This
year, for the first time, ITEC will be
joining forces with the American
Institute of Aeronautics and Astro-
nautics (AIAA)/IEEE Electric Aircraft
Technologies Symposium (EATS) for
one joint event, called ITEC +. The
Digital Object Identifier 10.1109/MELE.2021.3115545
Date of current version: 1 December 2021
The Westin Anaheim Resort exterior and the connected tourism district.
76 IE E E E l e c t r i f i c ati o n M agaz ine / DECEMBER 2021
chair, and Phillip Ansell, ITEC+ EATS
general cochair, will be the driving
forces behind ITEC+. Preindl has been
with ITEC for many years and offers
his thoughts on partnering with EATS:
Transportation electrification is
conference will continue to retain the a key thrust toward a secure,
elements that make ITEC unique, a connected, and sustainable
broad overview of transportation humanity. ITEC has emerged as
electrification, while highlighting the the premier conference with a
advancements in aerospace electrifi- global brand that serves the
cation and will be held 15–17 June transportation electrification
2022 at the Westin Anaheim Resort, industry and fosters industry
Anaheim, California. and academia interaction. We
In a truly collaborative effort, Mat- are excited to partner with AIAA
thias Preindl, ITEC+ EATS general and join forces with EATS to
 even better serve the communi-
ty and leverage synergies on the
path toward electrifying land,
sea, air, and space transportation.
Ansell, joining ITEC from EATS,
has served on the EATS conference
committee for several years as well.
Ansell commented:
The shift toward electrification
in the aviation sector has opened
up entirely new worlds of capa-
bilities for future aircraft, mak-
ing these systems more sustain-
able, operationally robust, and
less expensive to maintain and
operate. We are thrilled at the
opportunity to bring the aero-
nautics and transportation elec-
trification communities together
within the ITEC+ EATS model,
allowing both communities
to learn together and advance
what is possible.
The tracks for the upcoming
ITEC+ include the following:
 “Power Electronics and Motor
Drives”
 “Electric Machines and Actuators”
 “Electrical Energy Generation and
Storage: Battery, Fuel Cell, Solar,
Regeneration, and Others”
 “Superconducting and Cryogenic
Systems and Components”
 “System Dynamics, Modeling,
and Control”
 “Mission Operation: Energy, Cost,
Emissions, and Thermal Manage-
ment”
 “Connected and Autonomous
Vehicles, Smart Mobility, and
Vehicle Functional Security”
 “Electrified Powertrain Architec-
tures: Design, Integration, and
Optimization”
 “Vehicle Design and System
Optimization for Electrified
Cars, Heavy Duty, and Off-Road
Transportation”
 “Vehicle Design and System
Optimization for Electrified
Airplane, Space, Sea, and Under-
sea Transportation”
 “Rapid Prototyping, Real-Time
Simulation, HIL and SIL, Verifica-
tion, Validation, and Testing”
 “Battery and Hydrogen Infrastruc-
ture and Operations: Smart and
Microgrids, EV-Interacting Grids,
and Hydrogen Logistics”
 “Battery Chargers: Onboard,
Wireless, Fast, and Ultrafast”
 “Safety, Diagnostics, Reliability,
Failure and Fault Mode Protec-
tion, and EMI”
 “Codes, Certification, Standards,
Policies, and Regulations for
Transportation Electrification.”
The partnership between IEEE
and AIAA during this time of acceler-
ated growth of the electrification
industry could not be more appropri-
ate. There is much to look forward to
at this year’s conference. We hope to
see you there. For more details, visit
https://itec-conf.com/.

Get Involved
IEEE Smart Village helps remote
communities overcome hurdles
and build sustainable, green
solutions. We need your help to
serve more communities!

Pledge a recurring monthly

micro-payment of $20/month.
1,000 engineers will provide
power to a remote region.

Volunteer with IEEE Smart

Village.

LINGSHED, INDIA

For more information, visit

ieee-smart-village.org

IEEE SMART VILLAGE IS AN

Power a Village, Empower a Community IEEE FOUNDATION PRIORITY INITIATIVE
Digital Object Identifier 10.1109/MELE.2021.3126975

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 77

 VIEWPOINT
Scaling up to Long-Duration Storage
material costs have slowed the pace
of recent cost reductions, and the
long-term needs of the power grid
may require more duration than is
cost-effectively achievable with a
cell-type battery. To explore this
question, this article examines
trends from adjacent industries.
Examining Historic Parallels
In charting where the large-scale,
stationary storage industry may
journey, some historical parallels
may be helpful. Although disrup-
tions are to be expected in any
accelerating field, the manner in
which new technologies become
established can inform the develop-
ment of long-duration storage
today. What follows are encapsula-
tions of three other industries in
various stages of disruption: light-
ing, currently experiencing a transi-
tion; computing, in a state with a
dominant technology; and power
generation, in a state with no domi-
nant technology.
100
80
60
(%)
Building Exterior
Overall
40
Linear Fixture
Low/High Bay
Decorative
20
Connected Horticultural
Lighting Lighting
0
Figure 1. The 2018 installed adoption of LED lighting applications. (Source: U.S. Department
of Energy.)
78 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
(continued from page 84)
In 1992, William Nordhaus ana-
lyzed measures of wage growth over
time. He wanted to develop a stan-
dard measure of output that could
transcend “periods of major techno-
logical change.” In an interview with
National Public Radio, Nordhaus
explained the increase in terms of
light for a day’s work, highlighting
the process of disruptive innovation
over time.
For most of history, artificial light
was provided by burning some form
of wax, fat, or oil. A day’s work would
provide enough value for roughly
150  lmh from a candle, equivalent to
a couple of hours from an incandes-
cent light bulb. In the mid-1800s, ker-
osene and gas lighting arrived, and
provided an order-of-magnitude
improvement in the amount of light,
offering approximately 5 h for a day’s
work. The incandescent lamp, in turn,
replaced burned fuels, and by the
1990s, a day’s labor was enough for
roughly 20,000 h of lighting. Despite
some promising improvements with
Traffic Signals
Exit Signs
Refrigerator Case
Display
Parking Garage
Parking Lot
Street/Roadway
A-Type
Directional
fluorescent lighting, today’s lighting is
increasingly dominated by solid-state
lighting or LEDs, which light up for
50,000 h—enough for a lifetime—
under the same metric. According to
an August 2020 U.S. Department of
Energy (DOE) lighting report, LEDs
reached 30% of the overall lighting
market by 2018, and as shown in Fig-
ure 1, are approaching 100% adoption
in key applications such as traffic sig-
nals and exit signs.
As with lighting, modern comput-
ing has experienced multiple “eras”—
p i o n e e r e d by v a c u u m t u b e s ,
subsequently displaced by discrete
transistors, and now dominated by
silicon-based microprocessors. Jona-
than Koomey, with a team from Law-
rence Berkeley National Lab, Stanford
University, Microsoft, and Intel,
chronicled the advancement in com-
putation. They gauged their progress
by measuring the number of calcula-
tions achievable per unit of electrici-
ty. In 1946, the first programmable
computer, an electronic numerical
integrator and computer (ENIAC),
provided 437 computations/kWh. By
2009, a consumer-grade multicore
processor delivered nearly a quadril-
lion calculations per kilowatthour.
Despite the somewhat slower
growth of the 1960s discrete transis-
tor era, the rate-of-efficiency gains
from ENIAC through modern PCs
was fairly constant, doubling roughly
every 1.6 yr. Figure 2 depicts how
each new generation of computing
technology fully displaced the previ-
ous generation.
In contrast with silicon-dominat-
ed computing, U.S. power generation
provides an example of a sector that
has reached a state with multiple
technologies. An internal DOE
analysis adapted the Herfindahl–
Hirschman index (HHI), primarily a
measure of market concentration, to
market “shares” of various fuels. By
modeling all the generators of a
given fuel type as a single “firm,” this
analysis sought to gauge how the
concentration of fuels has evolved
over time (Figure 3). Although coal
was the majority of U.S. power gener-
ation in the earlier part of the 20th
century, the emergence of nuclear
(1970s), natural gas (2000s), and now
renewables have entered the mix in
significant numbers. Measured by
HHI, market concentration has fallen
across the United States since the
1980s, indicating increasing diversity.
These three industries provide
varying analogies for how the storage
industry might incorporate a new
disruptive technology. In the case of
computation, each successive tech-
nology era—vacuum, transistor, and
microprocessor—fully supplanted
the previous one (although there
were many examples of hybrid vacu-
um-transistor computers during the
transition). On the other end of the
1.0E+16
1.0E+15
1.0E+14
1.0E+13
1.0E+12
1.0E+11
1.0E+10
Computations/kWh
1.0E+09
1.0E+08
1.0E+07
1.0E+06 IBM 704
1.0E+05
1.0E+04
Univac I
1.0E+03
ENIAC
1.0E+02
1.0E+01
1.0E+00
1940 1950
Figure 2. Over time, transistors and semiconductors wholly replaced previous computational architectures. (Source: Koomey et al.)
spectrum, power generation—the
technology mix—is expected to con-
tinue its increasing diversity. In
many cases, the diversity was pur-
posefully designed to mitigate fuel
price or weather volatility. Finally, in
2021, the lighting industry is in a
state of transition. Although the
underlying cost and performance
properties of LEDs appear to bright-
en their long-term future, other tech-
nologies, like fluorescent and
incandescent, remain economically
competitive or aesthetically attrac-
tive for specific applications.
Shaping the Storehouse
Like lumens for lighting, computa-
tional operations for processors, or
watthours for generators, all storage
devices perform a discrete function:
moving electricity from one time
period to another. As a part of the
Long Duration Energy Storage
Computational Efficiency and Architecture
Compaq iPaq
Compaq Deskpro
386/20e
Apple Macintosh
IBM PC
Compaq Deskpro
IBM PC/XT
Apple IIe
SDS 920 Commodore 64
UNIVAC 1107
DEC PDP-8
UNIVAC III
NCR 304
Univac II
Burroughs 204
EDVAC
1960 1970 1980 1990
Year
IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1
Earthshot, the DOE has adopted the
use of the metric-kilowatthour cycle,
meaning delivering 1 kWh through
the storage device. This unit is used
in the underlying Storage Earthshot
target of US$.05/kWh cycle levelized
cost of storage, or that the total cost
of storage resource, when levelized
over a lifetime of operations, should
be available to charge and discharge
a kilowatthour for less than a nickel.
Also like lighting, computation, or
generation, not all kilowatthour
cycles live in the same context. Cer-
tain applications may require spe-
cialized requirements, such as
transportation durability, very low
maintenance, or aesthetics. When
determining the levelized cost of stor-
age, two of the most important fac-
tors are the duration and duty cycle.
Over the next decade, the pre-
dominant expectation for grid stor-
age applications will be to facilitate
Dell Optiplex GX400
SUN Blade 1000
IBM PS/2 E
Vacuum
Hybrid Vacuum Transistor
Transistor
Semiconductor
2000 2010 2020
79
 VIEWPOINT

greater amounts of renewable energy.
Both the overall load shape and the
fastest-growing generation type—
solar photovoltaic—exhibit predomi-
nantly diurnal patterns. Storage
technologies with the capability to
scale to 10 h or more will provide
the most direct benefit in the medi-
um-term, mostly decarbonized
power grid. Beyond grid decarbon-
ization, economywide decarboniza-
tion may drive the adoption of
ultralong-duration storage, in the
100–1,000-h range. As seasonal loads
migrate to the electric power sys-
tem, the importance of seasonal
storage will increase.
Independent of decarbonization
trends, storage is also finding a role
in resilience applications. although
the financial viability of storage
requires substantially decreasing
cost as durations increase, the
value of a backup source of power
increases with the duration of an
outage. As documented in a survey
conducted by a Carnegie Mellon Uni-
versity team, consumers would be
willing to pay a significant premium,
as high as US$1.2/kWh, to maintain
limited services during an extended,
multiday outage.
Such events appear to be increas-
ing in both frequency and magni-
tude. For example, in 2009, an ice
storm in Missouri and Kentucky left
some customers without power for
weeks. The U.S. National Guard trav-
eled door to door with food and
emergency generators. During the
2021 winter event in Texas, hourly
electricity prices cleared at the mar-
ket cap of US$9,000/MWh for nearly
four days. Emergency load shed pro-
cedures were in place for 105 h.
According to the Energy Information
Administration, outages due to major
events have constituted the majority
of customer-interruption hours (as
opposed to outages due to power
system malfunctions) in every year
since 2016.
Meeting the Candidates
Given the diversity of storage appli-
cations, the final step is to examine
the technology candidates that have
a viable pathway to achieving the
required cost and performance tar-
gets. Roughly speaking, the candi-
dates can inhabit two families.
The first family, termed high per-
formance, low cost, maintains existing
performance characteristics while
significantly reducing costs. This
family would likely provide the daily-
to-weekly durations required for the
first stage of grid decarbonization as
well as basic resilience needs. The
performance ranges for this family
should exceed 70% round-trip effi-
ciency, and upfront capital costs

HHI Calculation by Generation Type

4.5 0.45

4 0.4

3.5 0.35
Million TWh Generated

3 0.3

2.5 0.25
HHI
2 0.2

1.5 0.15

1 0.1

0.5 0.05

0 0
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013

Year

Coal Petroleum Natural Gas

Other Gases Nuclear Hydro-Pumped Storage
Conventional Hydro Wood Waste
Geothermal Solar Wind
HHI

Figure 3. Generation diversity (measured by fuel type) has increased over time, with no currently dominant technology.

80 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021

 Nominate a
Colleague
Today!
For 2023 IEEE
Fellows Class!

ieee-pes.org/OAJPE

Submissions are due

1 March 2022

Digital Object Identifier 10.1109/MELE.2021.3126977

 VIEWPOINT
should approach US$30–US$40/kWh.
The acceptable degradation over
3,000–5,000 cycles would enable a
reasonable 10–15-yr service life from
daily cycling.
The second family, termed moder-
ate performance, ultra-low cost, trades
off some performance characteristics
with dramatically lower costs. This
family would enable the weekly-to-
seasonal durations required for the
second stage of grid decarbonization
as well as the most demanding resil-
ience needs. The performance
ranges for this family could have
round-trip efficiencies of approxi-
mately 50% or slightly less, which
would be offset by upfront capital
costs below US$10/kWh. Given that
these resources would only occa-
sionally be cycled, acceptable degra-
dation over hundreds or dozens of
cycles could still enable a sufficiently
long service life.
Any new technology would need
to overcome the advantages of the
current dominant technology: Li-ion.
For Li-ion batteries in 2021, their
long-term performance is well char-
acterized, the supply chain is devel-
oped and expanding, and their costs
continue to decline (albeit at a slow-
er pace). Absent external interven-
tion, performance and supply chains
take time to establish. Therefore, the
most promising path for a new tech-
nology to displace Li-ion is to dem-
onstrate a significant cost advantage
for a given application.
High Performance, Low Cost
The technologies in the high-perfor-
mance category are likely to favor
those that store energy in higher-
quality forms, such as mechanical or
electrochemical. Chemical or
mechanical energy can be readily
converted to and from electricity at
relatively high efficiencies, which can
improve the cost profile for systems
that cycle frequently.
Pumped storage is the most well-
established gravitational-mechanical
system. Gravitational systems that
82 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
use pulleys and railcars have also
been explored. In many cases, these
mechanical systems incorporate
exceptionally low-cost storage media
(like rocks or water) but have conver-
sion equipment or geographic
deployment constraints that can
drive up costs elsewhere in the sys-
tem. Although these systems may
have competitive levelized costs,
many markets today lack the struc-
tures to invest in an asset with a life-
time of 50-plus years.
The other major class of high-per-
formance technologies include
emerging electrochemical batteries.
Most of these candidates use a flow-
battery architecture (as opposed to
individually packaged cells), where
energy and power can be scaled
independently. Innovators are forging
an entire new class of electrochemis-
tries that use low-cost materials like
iron, zinc, or Na, bypassing the Li-ion
supply chain entirely. Theoretically,
most of these chemistries have
attractive cost profiles, although the
material cost of containment (i.e.,
tanks) may establish an upper bound
in the tens of hours for which these
technologies would be cost-effective.
A few of these technologies are cur-
rently manufactured at the scale
that would maximize their cost-
reduction potential.
Moderate Performance,
Ultralow Cost
The technologies in the ultralow-
cost category are likely to favor stor-
age mediums that utilize lower-
quality energy forms, i.e., heat.
These technologies can use very
low-cost materials as the storage
medium, such as concrete or mol-
ten salt. Although some technolo-
gies were envisioned to use heat
from concentrating solar or other
sources, their very low costs enable
grid electricity a viable input as a
heat source. After being stored in an
insulated medium or through an
endothermic reaction, thermal
energy typically requires a heat
engine for conversion back to elec-
tricity, which limits its electricity-
only efficiency. To be suitable for
very long-duration applications,
innovators are examining methods
to best insulate the heat, reducing
self-discharge over time.
Because these technologies have
low, round-trip efficiencies, their
optimal competitive use may be for
longer (beyond tens of hours) dura-
tions, where other technologies
become impractical or too expensive.
Some of these technologies still
require earlier-stage R&D, especially
for innovations around materials.
The other technologies incorporate
largely off-the-shelf components. For
the most part, these technologies still
need to be demonstrated and manu-
factured at scale.
The Case for Diversity
To date, there does not appear to be
any candidate storage technology
that simultaneously exhibits every
desirable characteristic, including as
high efficiencies, high cycle life, low
cost, or geographic portability. There-
fore, it is difficult to envision that any
technology in existence or develop-
ment could follow a pathway to
dominance that semiconductors or
LEDs exhibit in their respective
domains. A more likely outcome is
that a diversity of storage technolo-
gies will coexist to serve the variety
of energy-balancing needs, from dis-
tributed to centralized architectures,
and from hourly durations to weekly
or even seasonal.
Concluding Currents
In some ways, the future of storage as
a perpetually diverse industry is
already here. For example, research-
ers and commercial vendors are
already piloting hybrid pumped
hydro and battery systems, where the
battery reduces mechanical stresses
by offloading very short-term varia-
tions. Although other industries like
(continued on page 83)
AD INDEX & SALES OFFICES

The Advertisers Index contained in this issue is compiled as a service to our readers and advertisers: the
publisher is not liable for errors or omissions although every effort is made to ensure its accuracy. Be
sure to let our advertisers know you found them through IEEE Electrification Magazine.

C O M PA N Y PAG E URL PHONE

GMW Associates Cover 4 gmw.com/ds
RTDS Technologies 5 rtds.com

IEEE Electrification Magazine Representative

Erik Henson
Naylor Association Solutions
Tel: +1 352 333 3443
Fax: +1 352 331 3525
ehenson@naylor.com

Digital Object Identifier 10.1109/MELE.2021.3129826

Scaling up to Long-Duration Storage (continued from page 82)

computation have shown that a new
technology can come to dominate a
field, this pattern appears unlikely in
energy storage. Despite the recent
overwhelming market share of Li-ion-
based energy storage systems, the
future long-duration needs of the
decarbonized grid are likely to exceed
the remaining cost-reduction poten-
tial for Li. And even though a variety of
technologies show promising cost and
performance characteristics, no single
technology appears to have compel-
ling metrics in every category. Given
the variations in both uses and tech-
nologies, the long-term future of the
energy storage industry is likely to
mirror that of the grid itself—with a
healthy diversity of complementary
technologies, all working in concert to
enable a safe, reliable, economical, and
clean power system.
Disclaimer
The views and opinions of the
author expressed herein do not nec-
essarily state those of the United
States Government or any agency
thereof.
For Further Reading
S. Baik, S. Sirinterlikci, J. W. Park, A. Davis,
and M. Granger Morgan, “Estimating res-
idential customers’ costs of large, long-
duration blackouts,” in Proc. Front. Econ.
Widespread, Long-Duration Power Interrup-
tions, Expert Workshop, Jan. 4, 2019, p. 1.
J. G. Koomey, S. Berard, M. Sanchez,
and H. Wong, “Implicatoins of histori-
cal trends in the electrical efficiency
of computing,” IEEE Ann. H i s t o r y
Comput., vol. 33, pp. 46–54, July/Sept.
2011.
W. D. Nordhaus, “Do real-output
and real-wage measures capture reali-
ty? The history of lighting suggests
not,” in The Economics of New Goods, T. F.
Bresnahan and R. J. Gordon, Eds. Chica-
go: Univ. of Chicago Press, Jan. 1996,
pp. 27–70.
Biography
Eric Hsieh (eric.hsieh@hq.doe.gov) is
with the U.S. Department of Energy,
Washington, D.C. 20585, USA.

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 1 83

 VIEWPOINT
Scaling up to Long-Duration
Storage
By Eric Hsieh
ODAY’S LITHIUM (LI)-
T based battery has ushered
in an era of cost-effective,
gigawatt-hour-scale storage resources.
However, the future of Li as the domi-
nant or even preferred technology is
uncertain. History is replete with
examples of pioneering technologies
being displaced by subsequent inno-
vations, including lighting, computa-
tion, and power generation. Given the
fundamental properties of candidate
technologies and target markets, this
article identifies a diverse technology
portfolio as the most likely future for
the storage industry.
The Winding Path of
Large-Scale Storage
Pumped storage was the first mega-
watt-scale storage technology on the
early U.S. grid, with a 31-MW plant in
Connecticut that began operations
between 1928 and 1930. The combina-
tion of high efficiency and low cost
allowed the technology to become
widespread in a few decades. Of the
21.9-GW, operational pumped-storage
capacity today, 14.5 GW, or roughly two-
thirds, was built between 1960 and
1980. Today, pumped storage still occu-
pies the top of the charts in all storage-
deployment metrics, except for growth.
Before turning to the new fast-
est-growing storage technology, it’s
Digital Object Identifier 10.1109/MELE.2021.3115546
Date of current version: 1 December 2021
84 I E E E E l e c t r i f i cati o n M agaz ine / DECEMBER 2021
worth examining the other technol-
ogies encountered on the storage-
adoption path. Sodium sulfide
(Na2S) and lead acid were examples
of early second-generation pioneers.
Some selected initial projects dem-
onstrated the technology’s feasibili-
ty, such as the 20-MW, 14-MWh
lead-acid battery installed in 1994
by the Puerto Rico Electric Power
Authority. An early Na2S battery was
installed by American Electric Power
(AEP) Ohio in 2002. However, the
maintenance or operations require-
ments of these battery technologies
relegated further deployments to
niche applications.
The AEP site also tested a fly-
wheel-based storage system. Subse-
quent demonstrations showed that
flywheels were particularly suited to
short-duration, high-power applica-
tions, like frequency regulation. With
the potential for a full charge-dis-
charge cycle every hour, a storage
resource providing frequency regula-
tion might have to undergo thou-
sands of cycles every year. The
accurate performance and long life-
time indicated a potentially viable
commercial business case for fly-
wheels. The test results convinced
regional markets and federal regula-
tors to revise compensation rules in
such a way that would recognize the
incremental value of fast-responding
devices, like flywheels and batteries.
Unfortunately for champions of
flywheels, a lower-cost competitor
was already on deck.
In many ways, Li-ion batteries
were the beneficiaries of doubly
lucky timing. Although flywheels
were demonstrating how new stor-
age technologies could provide grid
benefits, Li-battery manufacturing
benefited from investments in elec-
tric vehicle manufacturing. With cost
reductions exceeding analyst projec-
tions every year, Li-based energy
storage systems quickly displaced
flywheels as the least-expensive
technology for frequency regulation,
even after accounting for the longer
lifetimes of flywheels. As Li-ion pric-
es continued their nearly 90% price
decline through the 2010s, the tech-
nology not only displaced other stor-
age technologies but also began to
displace peaking and midmerit ther-
mal units. According to IHS Markit
(formerly known as Information Han-
dling Services), the United States saw
1.7 GW and 3.8 GWh of new battery
storage capacity installed in 2020. Of
this amount, all but 30 MWh were Li
ion. The expectations for 2021 instal-
lations exceed 5 GW and 14 GWh,
with more than 99% based on Li.
Given the overwhelming market
share of Li-ion batteries, a keen
observer may ask how long this
dominance may endure. Supply-
chain constraints and underlying
(continued on page 78)
2325-5897/21©2021IEEE
 POWER IN
CONNECTIONS
APRIL 25-28, 2022 | NEW ORLEANS

The 2022 IEEE PES T&D Conference & Exposition returns this April, reuniting the
industry to explore topics driving the future of energy. Join us to engage in critical
conversations on preparing for the challenges ahead.

REGISTRATION OPENS JANUARY 2022.

VISIT IEEET-D.ORG OR SCAN THE QR CODE TO LEARN MORE.

www.ieeet-d.org

Digital Object Identifier 10.1109/MELE.2021.3126978

HIGH ACCURACY
DC/AC ZER0-FLUX CURRENT TRANSDUCERS
DELIVERY FROM
STOCK UP TO 3 WEEKS
See current availablity at gmw.com/ds/

DANISENSE HIGH ACCURACY AND HIGH

RESOLUTION DC/AC ZER0-FLUX CURRENT
TRANSDUCERS
±12.5A to ±11000A, high accuracy Current
Transducers for Current Measurement
Applications: Power measurement and power
analysis, high accuracy current emulators or current
sources and sinks, gradient amplifiers for MRI devices,
precision motor drives, battery testing and evaluation
systems, and current calibration reference transducer.
• Very high absolute amplitude and phase accuracy
from dc to over 1kHz
• Low signal output noise and exceptional long term
stability
• Low fluxgate switching noise on the pimary
• Enhanced electrostatic shielding to increase rejection
of primary dV/dt coupling
• Increased operating temperature range
• Reduced mechanical dimensions
• Options: Current Output or Voltage Output Signal,
Calibration Winding
• Amplitude and Phase measurement to 300kHz
included with each head

Conveient, compact multi-channel systems

• DSSIU-4-1U and DSSIU-6-1U 4-channel and 6-channel
system interface units, power multiple current trans-
ducers from a single 1U rack mount unit.
• Conveniently mix different transducers, for example:
(3) ±600A and (1) ±2000A transducers.
• Supports calibration winding for in-situ calibration
check of compatible transducers.

WWW.GMW.COM | INSTRUMENTATION FOR ELECTRIC CURRENT AND MAGNETIC FIELD MEASUREMENT